<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Versent on Medium]]></title>
        <description><![CDATA[Stories by Versent on Medium]]></description>
        <link>https://medium.com/@versent?source=rss-36669c51ff0f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*eB8RomPkGTQ8MrF8Z0S8Jg.jpeg</url>
            <title>Stories by Versent on Medium</title>
            <link>https://medium.com/@versent?source=rss-36669c51ff0f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:23:16 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@versent/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[CloudWatch RUM + X-Ray]]></title>
            <link>https://versent.medium.com/cloudwatch-rum-x-ray-b8c38a81fee6?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b8c38a81fee6</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Thu, 26 Feb 2026 23:11:51 GMT</pubDate>
            <atom:updated>2026-02-26T23:11:51.123Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fYsVcSs5hN039rO1KYMqXQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/man-in-black-and-white-checkered-dress-shirt-sitting-on-black-office-rolling-chair-w_zE6qlkQKA">This is Engineering </a>on Unsplash</figcaption></figure><h4>AWS</h4><h4>The Underrated Duo for Cloud Native Observability</h4><p>By Shaun Clements</p><h3>Lightweight, Low Complexity Cloud Native Tracing</h3><blockquote><em>TL;DR: Native is always my goto path when evaluating what is the most simple and cost-effective approach. When it comes to Observability tracing for a web application, AWS Cloudwatch RUM &amp; AWS X-Ray when combined, offer really powerful insights into your platform services from user behaviour perspective.</em></blockquote><p>With modern application architectures distributed across numerous technologies and broken down into microservices, tracing is hard. When a user request from a website is invoked, a modern run and maintain team investigating an incident raised, will often want to know:</p><ul><li>when was the request invoked</li><li>what actions were performed across AWS services</li><li>how long did it take to process each request</li><li>were there any errors reported in the client (javascript) or from network request (http)</li><li>events including client interactions to view the sequence/parallel actions performed</li></ul><p>With Cloudwatch RUM enabled web clients, along with X-Ray enabled AWS Services you are able to view all the above. This blog aims to show you how easy it is to get started and uplift your current observability practices.</p><h3>Cloudwatch RUM</h3><p>CloudWatch RUM (Real User Monitoring) is an APM (Application Performance Monitoring) feature from AWS, introduced April 2021 under CloudWatch’s Application Signals. It helps developers understand how their web applications are performing from the perspective of actual users. It collects data on user interactions, page load times, errors, allowing teams to identify performance bottlenecks and improve the overall user experience.</p><p><strong>What it does:</strong></p><ul><li>Monitors real user interactions with web applications.</li><li>Provides insights into performance metrics such as page load times and error rates.</li><li>Helps in diagnosing issues affecting user experience.</li><li>Integrates with AWS X-Ray for end-to-end tracing</li></ul><h3>AWS X-Ray</h3><p>AWS X-Ray is a powerful service that helps developers trace, analyze and debug distributed applications, particularly those built using microservices architecture. It provides insights into the performance of applications by tracing requests as they travel through various services, allowing teams to pinpoint bottlenecks and troubleshoot issues effectively.</p><p><strong>What it does:</strong></p><ul><li>Traces requests across multiple services to provide a comprehensive view of application performance.</li><li>Visualizes service maps to identify dependencies and interactions between services.</li><li>Captures detailed information about requests, including latency and error rates.</li><li>Integrates seamlessly with other AWS services, enhancing observability and debugging capabilities.</li></ul><p>By utilizing AWS X-Ray, developers can gain a deeper understanding of their application’s behavior, leading to improved performance and user satisfaction.</p><p><em>Since authoring this blog, AWS have announced X-Ray is transitioning to OpenTelemetry (OTel) as its primary instrumentation standard for application tracing and observability. A future revision of the blog can be expected.</em></p><h3>How to get started</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*g02kg3iVvgYYZrYQ" /><figcaption>Figure 1. High Level Architecture</figcaption></figure><h3>1. Creating the Cloudwatch RUM ‘App Monitor’ in AWS Cloud</h3><p>Add a Cloudwatch RUM AppMonitor in your AWS account enabled for XRay<br>Note: There are a number of additional telemetries you may also want to enable.<br>E.g. Using CDK below you could capture telemetries such as Errors, HTTP requests and enable X-Ray trace capturing.</p><p><strong>new</strong> CfnAppMonitor(<strong>this</strong>, `RumAppMonitor-{envName}`, { domain: domainName, name: `RumAppMonitor-{envName}`, customEvents: { status: &quot;ENABLED&quot; }, appMonitorConfiguration: { allowCookies: true, enableXRay: true, sessionSampleRate: 1, telemetries: [&#39;errors&#39;, &#39;http&#39;] } })</p><p><em>ref: </em><a href="https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/CloudWatch-RUM-get-started-create-app-monitor.html"><em>Creating a CloudWatch RUM app monitor for a web application — Amazon CloudWatch</em></a></p><h3>2. Provision an Identity Provider and Role for signed in Web client users to use to push RUM events in your AWS Account</h3><p>In order for a web client to push RUM messages to your RUM App monitor, the authenticated web user will use their JWT token, and the STS client to assume a role in the AWS workload account.</p><p>This particular requirement isn’t well documented and involves creating an OIDC Identity Provider along with a relevant role which is assumed by the FederatedPrincipal.<br>E.g. using CDK</p><p><strong>const</strong> oidcProvider = <strong>new</strong> aws_iam.OpenIdConnectProvider(<strong>this</strong>, &#39;oidcProvider&#39;, { url: <em>//issuer url for the Identity Provider,</em> clientIds: [&quot;XXXX-XXXX-XXXX-XXXX&quot;] <em>// registered idp client identifier</em> } <strong>const</strong> role = <strong>new</strong> aws_iam.Role(<strong>this</strong>, &#39;rumAssumeRole&#39;, { assumedBy: <strong>new</strong> aws_iam.FederatedPrincipal( oidcProvider.openIdConnectProviderArn, { StringEquals: { [`issuer url:aud`]: &quot;XXXX-XXXX-XXXX-XXXX&quot;<em>//clientId</em> } }, &#39;sts:AssumeRoleWithWebIdentity&#39;, ), inlinePolicies: { rumPolicy: <strong>new</strong> aws_iam.PolicyDocument({ statements: [ <strong>new</strong> aws_iam.PolicyStatement({ effect: awsm_iam.Effect.ALLOW, actions: [&#39;rum:PutRumEvents&#39;], resources: [`//arn of Rum App Monitor`] }) ] }}) } ) }})</p><h3>3. Add the RUM SDK Client to your React web application</h3><p>Start with creating and configuring your Cloudwatch RUM client in your web application ideally using the React Provider pattern.</p><p>Wrapping your React pages with the Rum Provider will automatically capture the telemetries specified across child pages with the option to still manually record custom events. E.g. Using npm package: <a href="https://www.npmjs.com/package/aws-rum-web">https://www.npmjs.com/package/aws-rum-web</a></p><p><strong>const</strong> rumClient = <strong>new</strong> AwsRum(&#39;7b85c38a-3518-4047-9e1b-438e3a75531b&#39;, <em>//Guid of the AppMonitor provisioned in your AWS account</em> &#39;1.0.0&#39;, region, { sessionSampleRate: 1, endpoint: `https://dataplane.rum.${region}.amazonaws.com`, enableXRay: true, allowCookies: true, telemetries: [ [&#39;errors&#39;, { stackTraceLength: 512}], [&#39;http&#39;, { stackTraceLength: 512, addXRayTraceIdHeader: true, recordAllRequests: true, urlsToExclude: [ <strong>new</strong> RegExp(`https://dataplane.rum.${region}.amazonaws.com/.+`) <em>//prevent RUM events traced in XRay</em> ] }] disableAutoPageView: true } }) <strong>const</strong> credentials = stsClient.send( <strong>new</strong> AssumeRoleWithWebIdentityCommand({ RoleArn: &quot;&quot;, <em>//Arn of the rumAssumeRole</em> RoleSessionName: &quot;Rum Assume Role&quot; WebIdentityToken: &quot;&quot; <em>// IdentityToken of the signed in User</em> }) ) rumClient.setAwsCredentials(credentials) <em>// You&#39;re now good to start pushing Rum Events!</em> rumClient.recordEvent(&quot;My Custom Event&quot;, { timestamp: Date.now()})</p><p>The STS SDK Client does not currently support token renewal when it expires so its up to you to include renewal functionality in your web client code.</p><h3>4. Enable X-Ray for all your AWS supported services</h3><p>Turn on X-Ray support for all AWS supported services you wish to trace.</p><h3>What does the data actually look like?</h3><p>CloudWatch RUM provides several key data points:</p><h4>Performance Metrics</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*BkmfnHDOW-9kBLn7" /><figcaption>Figure 2. Performance</figcaption></figure><ul><li><strong>Page Load Time</strong>: Measures the time taken for a page to fully load, providing insights into how quickly users can interact with the application.</li><li><strong>First Contentful Paint (FCP)</strong>: Indicates the time it takes for the first piece of content to be rendered on the screen, which is crucial for user perception of speed.</li><li><strong>Time to Interactive (TTI)</strong>: Measures how long it takes for a page to become fully interactive, allowing users to engage with the application.</li><li><strong>Error Rates</strong>: Tracks the frequency of errors encountered by users, including JavaScript errors and HTTP request failures, helping teams identify and resolve issues quickly.</li><li><strong>User Interaction Metrics</strong>: Captures data on user interactions such as clicks, scrolls, and navigation events, providing a comprehensive view of user engagement.</li><li><strong>Session Duration</strong>: Measures the length of time users spend on the application, which can indicate user satisfaction and engagement levels.</li><li><strong>Geographic Distribution</strong>: Provides insights into where users are accessing the application from, helping teams understand performance variations across different regions.</li></ul><h4>Errors</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zcabDBbOCCmeSKRx" /><figcaption>Figure 3. Errors</figcaption></figure><p>CloudWatch RUM also provides detailed insights into errors that occur during user interactions. This includes:</p><ul><li><strong>JavaScript Errors</strong>: Captures any errors that occur in the client-side JavaScript code, allowing developers to identify and fix issues that may affect user experience.</li><li><strong>HTTP Request Failures</strong>: Monitors failed HTTP requests, providing information on the status codes returned and the context of the failures.</li><li><strong>Error Rate Trends</strong>: Analyzes trends in error rates over time, helping teams to understand if issues are becoming more frequent and to prioritize fixes accordingly.</li><li><strong>User Impact</strong>: Identifies which users are affected by specific errors, allowing teams to address critical issues that impact user satisfaction.</li></ul><h4>X-Ray Traces</h4><p>AWS X-Ray traces provide a detailed view of the requests made to your application, allowing developers to analyze and debug issues effectively. When integrated with CloudWatch RUM, X-Ray can enhance observability by correlating user interactions with backend service performance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*06m1FtN4V82XN8tU" /><figcaption>Figure 4. Trace Query including Client Request</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-UYGW3M7AU_p3q-z" /><figcaption>Figure 5. Trace segments timelines</figcaption></figure><h3>How X-Ray Traces Work with CloudWatch RUM</h3><ol><li><strong>Correlation of Data</strong>: When a user interacts with a web application monitored by CloudWatch RUM, the RUM service injects aX-Amzn-Trace-Id header and collects telemetry data. This header is crucial for correlating user actions with X-Ray traces, enabling a comprehensive view of the request lifecycle.</li><li><strong>Tracing Requests</strong>: As requests travel through various microservices, X-Ray captures detailed information about each segment of the request. This includes latency, errors, and service interactions, providing insights into where bottlenecks may occur.</li><li><strong>Visualizing Performance</strong>: X-Ray visualizes the flow of requests through your application, allowing teams to identify dependencies and interactions between services. This visualization helps in pinpointing performance issues and understanding the overall health of the application.</li><li><strong>Error Analysis</strong>: By analyzing X-Ray traces alongside CloudWatch RUM data, teams can identify which user interactions lead to errors and how those errors impact user experience. This dual perspective is invaluable for troubleshooting and improving application reliability.</li><li><strong>Enhanced Debugging</strong>: The integration of CloudWatch RUM and X-Ray allows developers to trace issues back to specific user actions, making it easier to diagnose problems and implement fixes. This leads to a more responsive and user-friendly application.</li></ol><h3>The bottom line</h3><p>Getting meaningful observability into your web application doesn’t have to mean adopting complex third-party tooling. By combining CloudWatch RUM and AWS X-Ray, you get a lightweight, native solution that connects real user behaviour directly to backend service performance — from a client-side JavaScript error all the way through to a downstream microservice call.</p><p>The four-step setup is straightforward, the data it surfaces is immediately actionable, and it integrates naturally into an existing AWS workload. If you’re running a modern web application on AWS and observability is still an afterthought, this is one of the lowest-effort, highest-value uplifts your team can make</p><p>ref: <a href="https://docs.aws.amazon.com/xray/latest/devguide/xray-services-cloudwatch.html">CloudWatch integration with X-Ray — AWS X-Ray</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b8c38a81fee6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Back to Vegas]]></title>
            <link>https://medium.com/versent-tech-blog/back-to-vegas-cf558a605299?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/cf558a605299</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Thu, 02 Oct 2025 02:46:23 GMT</pubDate>
            <atom:updated>2025-10-02T03:03:37.187Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QYXQ-xXsaVtPSDqBI_yExA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/a-sign-that-says-welcome-to-fabulous-las-vegas-nevada-YeEx_N0QcxQ">Samuel Branch</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h4>aws</h4><h4>My AWS re:Invent 2025 Game Plan</h4><p><em>by Jason Wood, Lead AWS Consultant</em></p><p>So, I’ll start this off by saying I’ve only been once before in 2019, and<br>it was an amazing experience. Here’s my take on what worked for me<br>last time, and what I’ll be doing differently in 2025.</p><p>Oh, and if you don’t know what I’m talking about… it’s AWS re:Invent<br>2025! I have the privilege of going this year as part of the Versent crew, repping Modern Run — our Managed Services offering!</p><p>While re:Invent is not until the start of December 2025, you need to plan early. There are a LOT of sessions — but they fill up super quickly. Being part of Versent Modern Run, my focus in 2025 will be a lot more on AWS cloud operations, but there will be a lot in this post for everyone.</p><h3>Session Types: An Overview</h3><p>First and foremost, this is a tech conference, so there is a plethora of sessions. For those of you who are happy with me using a word like plethora. Thanks! It means a lot.</p><p>What are these different session types? Deep breath. There are:</p><ul><li><strong>Chalk Talks: </strong>I missed these when I went last time, but they are first on my agenda this year. Interactive discussions with an AWS engineer. If you have a problem similar to what the topic is about, this format is a must. These are not recorded.</li><li><strong>Builders’ Sessions: </strong>Small groups led by AWS engineers with hands-on learning.</li><li><strong>Code Talks: </strong>A little similar to Chalk Talks, but code-focused.</li><li><strong>Workshops: </strong>I did a couple of these when I went and enjoyed them.</li><li><strong>Gamified learning, Self-paced training, Interactive training: </strong>These could be interesting if they align with your interests, given there will be AWS experts in the sessions to assist.</li><li><strong>Breakout sessions: </strong>These are your standard lecture-style sessions. These will be recorded, so you can watch them at your leisure away from re:Invent. Because of that, I generally look at other options. Some sessions will give you a chance for Q&amp;A, and you might get a chance to talk to the presenter, so that may be a good reason to go.</li><li><strong>Bootcamps: </strong>A full or half day. Given everything else you’d be missing, I’d only recommend these if you truly need them.</li><li><strong>AWS Certification exam prep: </strong>Another I wouldn’t recommend, unless you feel you need it. There are enough options outside re:Invent for this sort of thing.</li></ul><p>These sessions are generally only one-hour long, so in my experience you can’t expect to learn a lot — even the 300 and 400-level sessions. The levels indicate the complexity of the topics discussed. Expect to get enough information to whet your appetite and put you on the right path. There’s just not enough time for much else.</p><blockquote><strong><em>My TLDR recommendation:</em></strong><em> Chalk Talks, Code Talks, Workshops,<br>Builders’ Sessions.</em></blockquote><p>Also, the <a href="https://reinvent-%20planner.cloud/">Unofficial re:Invent Planner</a> is a god send.</p><h3>Allow travel time</h3><p>When planning your sessions, you’ll also need to take into account the distance between venues. The casinos themselves are huge! You need to allow plenty of time to get out of the hotel and onto the street or to the transport hub. If you have to transfer, I’d allow at least an hour. Due to these factors, my recommendation is to try to cluster your sessions within a hotel. Even if you are booked into a session, that is only valid until the start of the session time. If you’re late, you go to the back of the standby queue. Plan carefully.</p><h3>Don’t forget the Expo</h3><p>But what if you don’t get into a session, or you have some gaps? The Expo hall! Where do you get swag? The Expo hall! It will be massive and definitely worth a look. The vendors will also have talks and demos of their products. Attendance is often needed to get the best merch. Hopefully, there’ll also be decent espresso coffees.</p><h3>Keynotes — Worth it?</h3><p>What about the other experiences? I have been asked a fair bit about the keynotes. These are recorded and broadcast live. Is it worth going? If you’ve been before, maybe not. If this is your first time, then live the experience, baby! Monday Night Live is the only keynote that streams at an Aus/NZ-friendly time, and Pete DeSantis has always been entertaining. This year, we have Dave Brown hosting, so that will be interesting. Pete now joins Werner Vogels on Thursday, a keynote I never miss, even at stupid-o’clock in Aus. If you only go to one keynote, this is the one I recommend!</p><h3>All the other stuff</h3><p>There is soooooooo much more. The Sports Forum sounds awesome. The networking events are always worth it and a great place to meet people. re:Play goes off chop (That’s “wild” for non-Aussies)! This is the closing event and the perfect way to let your hair down after a long week. Then, there’s the Hallway Sessions. You go to re:Invent to learn, and sometimes, the best learning is talking to people in the hallway or waiting in line. If that discussion is really giving you value… ditch your session and keep chatting. You may just uncover a diamond.</p><p>re:Invent is a week of awesome! So much learning, but so much more than learning. If you’re going as part of a team, you make or strengthen relationships… something I’m really looking forward to with my Versent team. I’m also excited to hang out with Community Builders from other countries.</p><p>If you ever get a chance to go, jump on it! re:Invent is much more than a conference, it’s an incredible lifetime experience.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cf558a605299" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/back-to-vegas-cf558a605299">Back to Vegas</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Native Application Development on AWS using .NET]]></title>
            <link>https://medium.com/versent-tech-blog/native-application-development-on-aws-using-net-da9f871296eb?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/da9f871296eb</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Thu, 17 Jul 2025 06:54:47 GMT</pubDate>
            <atom:updated>2025-07-17T08:33:16.843Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y2fJWOfW6uK-7Yj2wn7CGA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/group-of-people-using-laptop-computer-QckxruozjRg">Annie Spratt</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Native Application Development on AWS using .NET</h3><h4>A brief overview of the maturity and support for the .NET platform on AWS</h4><p><em>By Scott Baldwin, Principal Solution Consultant &amp; AWS Ambassador</em></p><p>When most people think about native application development on AWS, they often think of the usual suspects: NodeJS, Python, Java, or even Go. Still, not many people realise the amount of work and extensive support AWS has for Microsoft’s .NET platform. While, like other languages, .NET has a well-supported SDK that abstracts the AWS resource APIs, it also has dedicated teams building extensions and tooling to make developing native cloud applications on AWS in .NET a productive and enjoyable experience. In this post, I want to highlight a few of these projects and discuss how AWS is investing heavily in ensuring .NET on AWS is a good choice.</p><h3>Lambda and Serverless</h3><h3>.NET Runtime Support</h3><p>Little-known fact: AWS started offering .NET Core Lambda runtime support as early as December 2016, significantly before Azure Functions shipped with the same functionality. The support for running .NET on AWS Lambda has been improving ever since. AWS even created the <a href="https://github.com/aws/aws-lambda-dotnet">.NET Lambda tool</a>, which integrates into the .NET CLI. This tool offers a range of useful commands for working with serverless applications based on the <a href="https://aws.amazon.com/serverless/sam">AWS Serverless Application Model (SAM)</a>, as well as several blueprints for creating various types of serverless projects.</p><p>Templates like “serverless.AspNetCoreWebApp” create a .NET Web API/MVC Web App fronted by an API Gateway and hosted in Lambda, with the ability to run/debug locally or deploy to AWS using `dotnet lambda deploy-serverless` to deploy the SAM template. There are many more blueprints for different patterns, which you can view after installing them with dotnet new -i “Amazon.Lambda.Templates:*” by using the dotnet new list-author --aws command.</p><h3>Power Tools</h3><p>To add robustness to your lambda-based applications, AWS has a set of <a href="https://docs.powertools.aws.dev/lambda/dotnet/">Power Tools for Lambda</a>. This provides high-level abstractions for crucial aspects of developing serverless applications, such as logging, tracing, and metrics, as well as abstractions for handling challenging tasks like ensuring Idempotency or batch processing of SQS or streaming data. This talk at AWS re:Invent, <a href="https://www.youtube.com/watch?v=OWBazBRsF2A">Build production-ready serverless .NET apps with AWS Lambda (XNT301)</a>, dives deep into some of the benefits of using these Power Tools.</p><h3>DynamoDB</h3><p>DynamoDB is a NoSQL database used for low-latency (single-digit millisecond) access to data for OLTP applications. While the AWS .NET SDK provides low-level API access to DynamoDB (as do all supported SDKs), the .NET SDK adds two higher-level abstractions. The <a href="https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/DotNetSDKMidLevel.html">document model</a> wraps some of the more repetitive and tedious work of interacting with DynamoDB API’s. The <a href="https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/DotNetSDKHighLevel.html">Object Persistence Model</a> goes even further and allows you to use high-level C# constructs to define your data model and read and write it to DynamoDB. For example, you can specify a class and use attributes as shown in the code snippet below:</p><pre>namespace orders.Models<br>{<br>    [DynamoDBTable(&quot;orders&quot;)]<br>    public class LineItem<br>    {<br>        [DynamoDBIgnore]<br>        public string OrderId<br>        {<br>            get;<br>            set;<br>        } = string.Empty;<br>        [DynamoDBHashKey(&quot;PK&quot;)]<br>        public string PartitionKey<br>        {<br>            get =&gt; $ &quot;O_{OrderId}&quot;;<br>            set<br>            {} // Setter needed for DynamoDB deserialization <br>        }<br>        [DynamoDBRangeKey(&quot;SK&quot;)]<br>        public string SortKey<br>        {<br>            get;<br>            set;<br>        }<br>        [DynamoDBProperty(&quot;product_name&quot;)]<br>        public string ProductName<br>        {<br>            get;<br>            set;<br>        } = string.Empty;<br>        [DynamoDBProperty(&quot;cost&quot;)]<br>        public decimal Cost<br>        {<br>            get;<br>            set;<br>        }<br>        [DynamoDBProperty(&quot;quantity&quot;)]<br>        public int Quantity<br>        {<br>            get;<br>            set;<br>        }<br>        [DynamoDBProperty(&quot;total&quot;)]<br>        public decimal Total<br>        {<br>            get;<br>            set;<br>        }<br>        private int _lineItemNumber;<br>        [DynamoDBProperty(&quot;line_item_num&quot;)]<br>        public int LineItemNumber<br>        {<br>            get =&gt; _lineItemNumber;<br>            set<br>            {<br>                _lineItemNumber = value;<br>                SortKey = $ &quot;LI_{value}&quot;;<br>            }<br>        }<br>        [DynamoDBProperty(&quot;order_id&quot;)]<br>        public string LineItemOrderId<br>        {<br>            get =&gt; OrderId;<br>            set =&gt; OrderId = value;<br>        }<br>    }<br>}</pre><p>Reading all the LineItems then becomes as simple as creating a DynamoDBContext and querying it like so:</p><pre>public async Task&lt;List&lt;LineItem&gt;&gt; GetOrderLineItemsAsync(string orderId)<br>{<br>    var partitionKey = $ &quot;O_{orderId}&quot;;<br>    var queryConfig = new DynamoDBOperationConfig<br>    {<br>        QueryFilter = new List &lt;ScanCondition&gt;<br>        {<br>            new ScanCondition(&quot;SortKey&quot;, ScanOperator.BeginsWith, &quot;LI_&quot;)<br>        }<br>    };<br>    var search = _dynamoDBContext.QueryAsync&lt;LineItem&gt;(partitionKey, queryConfig);<br>    return await search.GetRemainingAsync();<br>}</pre><p>Similarly, the DynamoDBContext can be used to store your data like so:</p><pre>public async Task SaveLineItemAsync(LineItem lineItem)<br>{<br>    await _dynamoDBContext.SaveAsync(lineItem);<br>}</pre><p>This eliminates much of the work involved in interacting with DynamoDB.</p><h3>Modernisation</h3><p>To truly leverage native application development on AWS, such as containers or Lambda, you need to be running the modern cross-platform version of .NET, rather than the legacy .NET Framework. Migrating from .NET Framework to .NET is not always a simple task, and the cost of doing so is often a blocker to organisations with legacy .NET Framework applications.</p><p>AWS recently released <a href="https://aws.amazon.com/transform/net/">AWS Transform for .NET</a>, which utilises an agent-based AI approach to facilitate modernisation efforts. This can speed up a modernisation project by up to four times. Once complete, you can then run your code on Linux and significantly reduce your hosting costs. While this isn’t the only thing required to modernize a legacy application for cloud-native development, it is a mandatory first step.</p><h3>.NET Aspire</h3><p>To simplify some of the more challenging aspects of developing distributed applications on .NET, <a href="https://learn.microsoft.com/en-us/dotnet/aspire/get-started/aspire-overview">Microsoft has introduced .NET Aspire</a>. This enables developers to have a more fluid local development and debugging flow, thereby increasing productivity when building distributed applications. AWS is actively working on <a href="https://github.com/aws/integrations-on-dotnet-aspire-for-aws">a set of integrations</a> to support the .NET Aspire hosting model. Currently, this includes being able to generate cloud resources using CloudFormation or CDK, utilising DynamoDBLocal in place of a DynamoDB table, and they are working on adding a Lambda integration to support the local running and debugging of .NET Lambda functions.</p><h3>Conclusion</h3><p>When AWS released its CDK, C# was among the supported languages, meaning that your entire AWS project, including the infrastructure it runs on, can be written using C#.NET. Also, important to note that most of these projects are open-source and have a strong community around them. It’s safe to say that .NET is very well supported on the AWS platform, and AWS is investing heavily in making it a compelling choice for application development.</p><p>At Versent, we have digital engineers with deep experience in developing native cloud-based applications on .NET (as well as all the usual suspects). We have delivered many high-quality projects utilising the .NET tools that AWS provides.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=da9f871296eb" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/native-application-development-on-aws-using-net-da9f871296eb">Native Application Development on AWS using .NET</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Embracing Wellbeing Through Movement]]></title>
            <link>https://medium.com/versent-tech-blog/work-life-d8ff669d129b?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/d8ff669d129b</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Tue, 03 Sep 2024 09:12:23 GMT</pubDate>
            <atom:updated>2024-09-03T09:16:16.938Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UCWY7Mh86u8EagU69upTpA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@maxvdo?utm_source=medium&amp;utm_medium=referral">Max van den Oetelaar</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h4>work-life</h4><h4>Insights from Our World</h4><p><em>By Dora Peake (dora.peake@versent.com.au)</em></p><h3>Our Commitment to Wellbeing</h3><p>At Versent, we recognise the importance of a holistic approach to well-being. We understand the interconnectedness of physical and mental health, a lesson that became especially clear during COVID-19. However, our commitment extends beyond the pandemic as we strive to foster a workplace where everyone can thrive.</p><p>Over the past year, I have been particularly proud of our partnership with <a href="https://www.pukaup.com/">PukaUp,</a> a company that champions a holistic approach to wellbeing, encompassing mindfulness, nutrition, exercise, and sleep. This partnership has been transformative for many of us, allowing us to reclaim our well-being with a focus on preventative measures rather than waiting until we are in crisis mode.</p><p>We’ve held several impactful workshops on critical areas such as sleep, nutrition, and movement. Each session has offered invaluable insights, and I have integrated many of the practical strategies recommended. One of my favourite sessions focused on movement and how it benefits not only our physical selves but also our mood, productivity, focus, stress levels, and sleep. The philosophy shared was simple yet meaningful: <strong>Make Movement Part of Your Day in Easy Ways</strong>.</p><h3>My Reflections</h3><p>In the world of consulting, where the pace can be relentless and the demands are high, maintaining well-being is often a challenge. However, it is precisely in these high-pressure environments that well-being becomes paramount. As a leader, my ability to perform at my best hinges on how well I take care of myself. One crucial aspect of this self-care is movement.</p><h3>The Importance of Movement</h3><blockquote>Movement is more than just exercise; it’s about integrating physical activity into our daily routines in ways that are both manageable and sustainable.</blockquote><p>Dr. Ashton Wilson emphasized that movement doesn’t have to be complex or time-consuming. It’s about finding easy ways to incorporate activity into our day. This could be as simple as taking the stairs instead of the elevator, standing up during phone calls, or taking short walking breaks throughout the day. These small changes can have a profound impact on our overall well-being.</p><h3>Physical and Mental Benefits</h3><p>For me, physical movement is the cornerstone of my daily routine, and the benefits have been astounding. Movement has helped clear my mind, increase my resilience to stress, and significantly reduce the risk of burnout. Here are some practical tips that I’ve found particularly effective:</p><h3>Practical Tips for Incorporating Movement — Tried and Tested</h3><ol><li><strong>Set Reminders</strong>: Technology is my friend. I use it to track my exercise and set reminders to stand up, stretch, and measure my movement daily.</li><li><strong>Active Meetings</strong>: Whenever possible, I opt for walking meetings. This helps me stay active and fosters more creative and dynamic discussions.</li><li><strong>Workspace Adjustments</strong>: I use a standing desk, which helps me alternate between sitting and standing throughout the day.</li><li><strong>Short Exercise Bursts</strong>: Between meetings, I integrate short bursts of exercise into my routine. A few minutes of stretching, star jumps, push-ups or bodyweight exercises completely re-energize me and break the monotony of prolonged sitting.</li><li><strong>Lunchtime Workouts</strong>: Whether it’s a quick brisk walk or an energizing Pilates class, I try to incorporate movement that sets me up for the afternoon, feeling refreshed and ready to tackle the rest of the day.</li></ol><h3>Movement During Times of Change</h3><p>Over the past nine months, we have undergone significant change and transition. Maintaining a routine of regular movement has become even more critical during this period.</p><blockquote>Change can bring uncertainty and stress, and staying physically active has helped me manage and navigate these challenges.</blockquote><p>Movement has served as a grounding practice, providing me with a sense of stability and control amidst the flux.</p><h3>Resources and Support</h3><p>I’m incredibly proud of our efforts to support well-being, both through our partnership with PukaUp and our Employee Assistance Program (EAP) provided through Converge International. This program is available to all our team members and their families and includes access to an app, a health insights library, employee and manager support, and a monthly digital health and wellbeing magazine.</p><h3>Conclusion</h3><p>I encourage everyone to embrace wellbeing and prioritize their health. It’s not just beneficial — it’s essential. With simple, consistent steps, we can all enjoy the profound benefits that come from staying active. Let’s make movement a part of our day, every day, and embrace the journey towards a healthier, more balanced.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d8ff669d129b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/work-life-d8ff669d129b">Embracing Wellbeing Through Movement</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS re:Invent 2023: Adam Selipsky’s Keynote Announcement Recap]]></title>
            <link>https://medium.com/versent-tech-blog/aws-re-invent-2023-adam-selipskys-keynote-announcement-recap-610ae537bae9?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/610ae537bae9</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Mon, 11 Dec 2023 23:08:15 GMT</pubDate>
            <atom:updated>2023-12-11T23:09:04.227Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Josh Lopez (josh Lopez@versent.com.au)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/925/1*at3n35yUILRTk7KqpcWT0g.jpeg" /></figure><p><em>Re:invent has well and truly started and it’s back to full swing which allows over 50,000 people in Los Vegas and 300,000 virtually to consume all the great content that’s produced during the week. I’m still super excited about all the awesome announcements that will be coming over the next days. The announcement that resonated with me the most in Adam Selipsky, CEO AWS keynote the most was:</em></p><ul><li><strong>Amazon Q </strong>its bringing GenAI to everyone and it’s integration across CodeWhisper and QuickSight is just the start of big things to come for this service. It’s exciting how easy it is so easily accessible in the AWS Console. It’s great to see all the GenAI advancements this year at re:invent.</li></ul><p>I’ve put together all the announcements from the keynote and we’re looking to get plenty more over the course of re:Invent so keep an eye out for future posts to get across all the announcements.</p><h3>EC2</h3><ul><li><strong>AWS Graviton4 </strong>is the most energy efficient chip that Amazon has built. It’s 40% faster for database applications and 45% faster for large Java applications. R8g instances for EC2 has been launched which is powered by AWS Graviton4.</li><li>AWS Trainium2 is 4x faster than AWS Trainium and is optimised for training foundation models with hundred’s of billions to trillions of parameters. It has 65 exaflops of on demand supercomputing performance.</li></ul><h3>Storage</h3><ul><li><strong>Amazon S3 Express One Zone </strong>is purpose built for accelerating data process and you can choose your AZ for the first time. It has single digit millisecond latency and saves up to 60% on compute cost and 50% lower costs vs S3 Standard.</li></ul><h3>GenAI</h3><ul><li><strong>Amazon Bedrock </strong>has seen multiple announcements at re:Invent.<strong> Fine tuning </strong>you can now point to models in S3 with context and can learn from your data to get tailored responses which is available Amazon Titan Text Lite, Cohere Command Line &amp; Meta Llama 2<strong>. Retrieval Augment Generation (RAG) with Knowledge Bases </strong>allows you to customise your model’s responses when you want it to consider new information for the model’s response. <strong>Continued Pre-training for Amazon Titan Text Lite &amp; Express </strong>this is a technique that uses unlabelled data to improve the model’s knowledge of your data<strong>. Agents for Amazon Bedrock </strong>allows for you to now configure agents to allow Foundation Models to take actions on your behalf. It can take multi step tasks across a workflow that you define.</li><li><strong>Guardrails for Amazon Bedrock </strong>which allows you to easily safeguard your applications by preventing harmful content through filtering and inline with your AI policies</li><li><strong>Amazon Q </strong>a new GenAI assistant to work for you at work. It lets you answer questions quickly and generate content using your data and your operations. Its built securely and with enterprise requirements at the forefront. It’s trained with over 17 years of Amazon knowledge. It can architect solutions and an expert the Well Architected Framework. It’s built and integrated into the AWS Console and CodeWhisper.</li><li><strong>Amazon Q — Code Transformation</strong> handles the complete upgrade of Java version in your applications. Amazon has migrated 1,000 applications in 2 days which is a phenomenal outcome. It will be supporting .NET in the future.</li><li><strong>Amazon Q is enabled in Amazon Quicksight (Preview) </strong>allows you to integrate Amazon Q with your Quicksight report building.</li></ul><h3>Data &amp; Analytics</h3><ul><li><strong>Zero ETL Integration with Amazon Redshift now supports Amazon Aurora PostgreSQL, Amazon RDS for MySQL and Amazon DynamoDB, Amazon OpenSearch Service </strong>this will remove the pain to build and integrate data pipelines to transfer the data into Amazon Redshift. It will allow you to do more with data.</li><li><strong>Amazon DataZone AI recommendations </strong>uses GenAI to create business descriptions for your data. It provides recommendations on how to use data for faster insights. It makes data immediately easier to understand and discover.</li></ul><h3>Call Centre Enablement</h3><ul><li><strong>Amazon Q in Amazon Connect </strong>gives agents a GenAI boost that will allow them to generate recommendations in real time on call. It will send suggestions in real time without any typing required.</li></ul><p>There’s plenty of announcements to digest that will accelerate your modernisation journey. If you need to speak to a consultant on how you can utilise these services to transform your business, please do get in contact with us.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=610ae537bae9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/aws-re-invent-2023-adam-selipskys-keynote-announcement-recap-610ae537bae9">AWS re:Invent 2023: Adam Selipsky’s Keynote Announcement Recap</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS Re:Invent 2023 — Werner Vogels Keynote]]></title>
            <link>https://medium.com/versent-tech-blog/aws-re-invent-2023-werner-vogels-keynote-4b5df756e8d5?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4b5df756e8d5</guid>
            <category><![CDATA[aws-reinvent-2023]]></category>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Mon, 11 Dec 2023 23:05:55 GMT</pubDate>
            <atom:updated>2023-12-11T23:05:55.454Z</atom:updated>
            <content:encoded><![CDATA[<h3>AWS Re:Invent 2023 — Werner Vogels Keynote</h3><p><em>By Justyn Green (justyn.green@versent.com.au)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/925/1*hIX29orTNITaHwdLLplMhQ.jpeg" /></figure><p><em>With the end of Re:Invent approaching, Dr. Werner Vogels hit the stage for the final keynote this morning to speak about the observations and interactions he has seen in recent months that has impacted him and how we can “all” be better and support our customers. With the incredible announcements being shared this week already, it was still amazing to hear that some of the best announcements were being saved until last.</em></p><p>With his can-do attitude and light-hearted stories, Dr. Vogels spoke about multiple topics that he is passionate and energetic about.</p><p>In this post, I’ll got over his topics and thoughts, whilst sharing some of the few, yet important announcements being released today.</p><h3>The Frugal Architect</h3><ul><li>Dr. Vogels kicked off his keynote with an allegory on the importance of architecture. Speaking openly around how cost drives the world we now find ourselves in.</li><li>The key theme was that we should all “architect with cost in mind,” and this concept was explored throughout the talk.</li></ul><h3>Dr. Vogels went on to discuss how there are 3 (THREE) themes in being a Frugal Architect.</h3><ol><li>Design</li><li>Measure</li><li>Optimise</li></ol><p>Whilst architecture an application includes services design, being clear about architecting for cost, there are 3 (THREE) principles that should be considered (&amp; regular reconsidered) in this process.</p><h3>Design</h3><p>Cost is a Non-Functional Requirement</p><ul><li>Whilst we continue to see cost as an after thought and a burden we review after a service has gone live, the focus should be directed to the left as part of the requirements.</li><li>Dr Vogels went on to discuss that as Architect’s partnering with the business to build requirements including cost targets must start at the beginning, not at the end.</li></ul><p>Systems that last align cost to a business</p><ul><li>Architecting systems that model the business profit centres to enable economies of scale.</li><li>Dr. Vogels demonstrated visually that when an application is built to provide a service that the scaling of customer demand should align to the costs being consumed so it doesn’t erode value.</li><li>But, building quickly, trade off’s are sometimes required. That’s okay, but it’s a choice where you will have to pay back those “technical debts”.</li></ul><p>Architecting is a series of Trade-Off’s</p><ul><li>It’s crucial to evaluate and re-evaluate technical and business trade-offs and invest in options that align to business needs.</li><li>Dr. Vogels shared a personal story on an Amazon service (Lambda) that they made a trade-off early, knowing that they had to pay back it’s technical debt.</li><li>Interestingly, when the choice was finally made, the first people to see the change were customers being pleasantly surprised by an increase in speed and stability.</li></ul><h3>Measure</h3><p>Unobserved systems lead to unknown costs</p><ul><li>Whilst investing time and effort (&amp; technology) upfront on monitoring and observability solutions can be costly, it is crucial to identify wasteful practices, painful workflows and strategical align resources for better experiences.</li><li>Dr. Vogels passionately shared a story about a customer that had poor experience with their systems until they realised that a large portion of their issues weren’t the application but their business logic inside their application that could be fixed with update in their code of 3x extra lines. He finished this story by saying, “know your costs”.</li></ul><p>Cost Aware architectures implement cost controls</p><ul><li>Taking action, you can implement fine grain controls that allow for cost optimisation and user experience.</li><li>Dr. Vogels shared a story he found interesting a few years ago with Amazon.com where he noticed that when poor user experience was impacted with the <strong><em>review </em></strong>tiles under a product starting to error, that purchase commitment and checkout efficiency dramatically decreased because customers value the input and opinions of their peers.</li><li>With tuneable architecture, businesses can enable or disable services to ensure optimal user experience.</li><li>Tiers of service types inform trade-offs and operations so that choices can be made to maximise customer satisfaction.</li></ul><h3>Optimise</h3><p>Cost Optimisation is Incremental</p><ul><li>Achieving cost savings is a journey that requires commitment and regular implementations reflected from the architecture of cost focus.</li><li>Dr. Vogels fondly shared the most famous story about Jeff Bezos business mantra, regarding the “fly-wheel” effect. Focus on always iterating on an objective to customer success to increase traffic and experience, to drive more traffic and better experience.</li></ul><p>Unchallenged Success leads to assumptions</p><ul><li>Always challenge the ways that things are done. As <em>Grace Hopper </em>once said, “the most dangerous thing we could say, is we’ve always done it this way.”</li><li>Dr. Vogels expressed deep passion for challenging our ways of architecture and to always scrutinise these technologies with a focus on trade-offs, cost optimisation and user experience. Never forget who we’re working for, (our customers), and become comfortable with disconfirming your beliefs.</li></ul><p>Personally, hearing these well articulated laws of architecture that Dr. Vogels discussed openly, it led me to wonder why these were not principles well defined as the dawn of computing or even cloud computing began. They’re simple, but powerful in enforcing a mindset on building with a team focus, and a way to innovate within a fast technology world that is to benefit the human race.</p><p>Whilst, the stories and principle’s shared almost felt like I was sitting in a technology philosophy session, excitedly there were several related AWS service announcements that will help adopt and easily adhere to these principles from “inception” to “reality”.</p><h3>Amazon CloudWatch Application Signals (Preview)</h3><p>With the removal of heavy lifting, soon to be released Amazon CloudWatch Application Signals allows for users to enable specific instrumentation against applications to increase visibility against the most critical business services. With an upcoming feature that allows for Service Level Objectives (SLO’s) to be configured, downstream notifications and alerts can be automatically raised when even the most minor component of an application landscape experiences an anomaly.</p><h3>Amazon SageMaker Studio: Code Editor (GA)</h3><p>Developers can now experience a more familiar and native approach to development of Machine Learning models with Amazon SageMaker Studio’s Code editor. Based on the OSS Code Editor for VS Code ©, this allows teams to use shortcuts, and integrations with their existing source code to write, run, debug and release code that natively operates with Amazon SageMaker machine learning capabilities and more.</p><h3>AWS myApplications (GA)</h3><p>With AWS myApplications, you are able to manage, monitor, orchestrate, secure, cost check and automate applications in your AWS environment.</p><p>The ability to have a single dashboard on a group of resources for a particular application, it unlocks the capability to see, manage and decide how to build, manage and improve applications from a known good baseline.</p><h3>Amazon Application Composer: Integrated Developer Environment for VS Code (GA)</h3><p>Finally, with a strong focus on enabling developers, unlocking capability by also adding direct IDE modules into VS Code for Application Composer, this service promises to allow developers to release at speed and scale no matter how trivial the project. Ensuring that the development lifecycle is simplified and accelerated by literally the click on 3x buttons.</p><h3>Learn and be curious:</h3><p>Before Dr. Vogels closed his presentation, he shared one more personal but empowering story.</p><p>And as you’d expect, it wouldn’t be without due expectations other than AI (Artificial Intelligence).Dr. Vogels wanted to have a go (how very Australian of him), with solving a real world problem using machine learning and trained models. Meeting with one of Dublin’s prestigious hospitals in their stroke department, he identified a relative simple problem that had yet to be solved. Detection of minor brain lesions with the use of image recognition and mass data processing and alerting.</p><p>Bring the story to a close, he shared that “if I can do it, you can do it!”</p><p>And with that final comment, he turned and said, “Now Go Build!”</p><p>If you need to speak to a consultant on how you can utilise some of the principle’s and new released services in AWS, to transform your business, please reach out.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4b5df756e8d5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/aws-re-invent-2023-werner-vogels-keynote-4b5df756e8d5">AWS Re:Invent 2023 — Werner Vogels Keynote</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS re:Invent 2023 — Data Analytics Keynote Announcements]]></title>
            <link>https://medium.com/versent-tech-blog/aws-re-invent-2023-data-analytics-keynote-announcements-0bdee2374f93?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/0bdee2374f93</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Mon, 11 Dec 2023 23:01:44 GMT</pubDate>
            <atom:updated>2023-12-11T23:19:58.234Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Josh Lopez (josh.lopez@versent.com.au)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/925/1*YmKqCsMLZZJUkSNFIuwW7g.jpeg" /></figure><p><em>Amazon’s Vice President of Database, Analytics and Machine Learning, Swami Sivasubramanian, delivered a fascinating keynote addressing the advancements in GenAI.</em></p><p>He focused on the advancements in GenAI over the last year and the symbolic relationship between data, GenAI and humans that’s accelerating transformation for customers. The announcement that resonated with me the most was:</p><ul><li><strong>Amazon Q data integration in AWS Glue </strong>is an integration which will significant increase the accessibility of data integration across AWS. It’s a service announcements that I know will have practical use cases for customers.</li></ul><p>I’ve put together all the announcements from the keynote and we’re looking to get plenty more over the course of re:invent so keep an eye out for future posts to get across all the announcements.</p><h3>GenAI</h3><ul><li><strong>Amazon Bedrock support for Anthropic Claude 2.1</strong> which reduces cost of prompts and completions on Bedrock. It has improved system prompts to improve the user experience.</li><li><strong>Model Evaluation on Amazon Bedrock </strong>it allows you to evaluate and compare the best foundational model for your use case. It will you to use curated datasets and predefined metrics for automatic evaluations.</li><li><strong>Amazon Bedrock support for Llama 2 70B </strong>which supports large scale tasks such as Language Models. It’s fine tuned for chat bots.</li><li><strong>Amazon Titan Multimodel Embeddings </strong>this model adapts to unique and proprietary data.</li><li><strong>Amazon Titan Text Lite</strong> is model are ideal for summarisation, copywriting and fine tuning.</li><li><strong>Amazon Titan Text Express </strong>is a model focused on open-ended text generation, conversational chat and supports RAG.</li><li><strong>Amazon Titan Image Generator</strong> which generates studio quality images using natural language prompts. It can be customised to proprietary data. It comes with an invisible watermark to support responsible AI.</li><li><strong>Custom model program for Anthropic Claude </strong>is a program will enable you to partner up with a team of experts to assist with customising the powerful models and securely deploy the models to Bedrock in your VPC.</li></ul><h3>Analytics</h3><ul><li><strong>Amazon SageMaker HyperPod </strong>is a new distributed capability that can reduce time to train models by up to 40%. It distributes the dataset across thousands of chips in a cluster and process it in parallel. It’s extremely resilient and will automatically replace any faulty compute. <strong>Amazon SageMaker Inference </strong>reduces model deployment by up to 50%. <strong>Amazon SageMaker Studio </strong>has a new user experience and capability.</li><li><strong>AWS Clean Rooms ML </strong>allows you to apply ML models with your partners without sharing the underlying data. It allows you to deploy the models in a few steps and saves months of development and resources using ready to apply ML models.</li></ul><h3>Database &amp; Search</h3><ul><li><strong>Vector engine for OpenSearch Serverless </strong>is a simple, scalable and high performing vector storage which allows for search to be operated without managing the underlying infrastructure</li><li><strong>Vector search capabilities now available in Amazon DocumentDB , Amazon MemoryDB for Redis and Amazon DynamoDB </strong>which provides single-digit millisecond query and update response times. Ten of thousands of queries per second with 99%+ recall</li><li><strong>Amazon Neptune Analytics </strong>is an analytics database engine for Amazon Neptune which makes it faster and easier to analyse amounts of graph data. It allows users to find insights from graph data up to 80x faster. Enables faster vector search by storing graph and vector data together.</li></ul><p>There’s plenty of announcements to digest that will accelerate drawing insights from your data. The innovations announced by <a href="https://www.youtube.com/watch?v=8clH7cbnIQw">Swami Sivasubramanian</a> are certainly going to accelerate the data analytics adoption of Versent customers as we start to implement these fantastic models and services. If you need to speak to a consultant on how you can utilise these services to transform your business, please do get in contact with us.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0bdee2374f93" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/aws-re-invent-2023-data-analytics-keynote-announcements-0bdee2374f93">AWS re:Invent 2023 — Data Analytics Keynote Announcements</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS re:Invent 2023 — the early bird needs to step up it’s game]]></title>
            <link>https://medium.com/versent-tech-blog/the-early-bird-needs-to-step-up-its-game-524fd2689286?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/524fd2689286</guid>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Mon, 11 Dec 2023 22:57:04 GMT</pubDate>
            <atom:updated>2023-12-11T23:20:58.032Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Jesse Aranki (jesse.aranki@versent.com.au)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VuuYFaJKXaJN41tBf7CTRg.jpeg" /></figure><p><em>Today marks the first day of re:Invent and the energy in the air is electric, and it would appear that I came into the day a lot less prepared than I thought. Walking through the Venetian you can almost drown in the sea of people trying to make their way to the morning’s first talks and workshops; and let me tell you — if you’re wanting to go to workshops you need to get here a lot earlier than you would think (lesson learned for tomorrow, for sure); before 9am and I have already felt the bitter string of rejection twice.</em></p><p>Arrive a least 45 mins ahead of the workshop if you don’t have your seats already reserved. That said however, there is so much going on at re:Invent that it’s hard to be too disappointed when there’s so much on offer## Low latency computing. Low latency computing is an interesting topic, as the demand for speed accelerates we are having to come up with new and interesting ways to approach and resolve the problems by demand. One of the early sessions i attended was a talk on delivering low latency apps by extending regions and AZ’s through the use of local zones and outposts.</p><p>For those unfamiliar a local zone can be thought of as a limited capability AZ that is generally deployed in large metro areas that aws does not already have an AZ presence in; such as Perth, Australia. Whilst they don’t offer the full suite of services that aws offers they are incredibly useful for delivering content where caching isn’t really an option (such as game servers) to your end users. For more information you can check out: <a href="https://aws.amazon.com/about-aws/global-infrastructure/localzones/">https://aws.amazon.com/about-aws/global-infrastructure/localzones/</a></p><p>Further to local zones, outposts is another way to tackle this problem; as you can quite literally house aws region grade server racks in your own data centre; which serves a few benefits</p><ul><li>Standardising your workflow and architecture across the board</li><li>You are able to essentially build your own private aws az should you choose.</li></ul><p>Looking at riot games for example, when optimising the game Valorant they were able to achieve extremely low latency to users across USA via the implementation of outposts, by using all AZs, local zones and then filling the gaps with outposts servers in areas where the distance between local zones and az’s were too large to satisfy their low latency goals you can check out the slides from a similar talk they had given back in 2021 if you’re interested to see the footnotes: <a href="https://d1.awsstatic.com/events/reinvent/2021/Riot_Games_at_the_edge_Launching_VALORANT_with_AWS_Outposts_GAM302.pdf">https://d1.awsstatic.com/events/reinvent/2021/Riot_Games_at_the_edge_Launching_VALORANT_with_AWS_Outposts_GAM302.pdf</a> ## monitoring serverless workloads.</p><p>If you haven’t entered into the wonderful world of serverless just yet, then this topic is one that’s really important to get a grasp of in the early phase of your journey. Not being very proficient with serverless I had some pretty good learnings coming out of this session around operational visibility in serverless workloads.Why is this so important? Because compute is ephemeral in the serverless world. No server no worries — unless of course you want to gain any insights into performance, or want to see the journey of data throughout your distributed system. Key Takeaways from the Session</p><ul><li>Understanding Ephemeral Compute: Since serverless functions are ephemeral, traditional monitoring methods don’t cut it. You need tools that can track and report in real-time.</li><li>Tracing Over Logging: While logging is useful, tracing provides a more comprehensive view of your system. It allows you to see the entire journey of a request across the system, not just what’s logged.</li><li>Tools and Techniques: The session introduced various tools and techniques for monitoring serverless applications. AWS X-Ray, for instance, is a powerful tool for tracing and analyzing serverless applications.</li><li>Real-World Scenarios: We saw examples of how companies have successfully implemented monitoring in their serverless environments. It’s one thing to talk about best practices, but seeing them in action is a whole different ballgame.</li></ul><p>As the day winds down, I can’t help but feel excited about what’s to come. There are still several days left of re:Invent, and if today was any indication, they’re going to be packed with learning, networking, and inspiration. If you’re here, let’s connect and share our experiences. And if you’re following along from afar, stay tuned for more updates. The early bird might need to step up its game, but trust me, it’s worth it!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uzr-llt0Pw0U-3tKsOmUjQ.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=524fd2689286" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/the-early-bird-needs-to-step-up-its-game-524fd2689286">AWS re:Invent 2023 — the early bird needs to step up it’s game</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Snowflake Data Governance Overview]]></title>
            <link>https://medium.com/versent-tech-blog/snowflake-data-governance-overview-9cdcd9468f5d?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/9cdcd9468f5d</guid>
            <category><![CDATA[snowflake]]></category>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Wed, 20 Sep 2023 05:38:40 GMT</pubDate>
            <atom:updated>2023-09-20T05:38:40.172Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Kovid Rathee (</em><a href="/service/mailto:kovid.rathee@versent.co"><em>kovid.rathee@versent.co</em></a><a href="http://m.au/">m.au</a>)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HD4teKDKamLJopEB" /><figcaption>Photo by <a href="https://unsplash.com/@dariuscotoi?utm_source=medium&amp;utm_medium=referral">Darius Cotoi</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Amidst a spree of data leaks over the last few years, data privacy and security have taken center stage in technology-related conversations. Securing the perimeter within and outside an organization has become extremely important. Networking constructs like VPCs, NACLs, and security groups provide a multi-layered approach to security. This, topped up with a solid identity and access management framework, does well with the application and infrastructure layer, but do all of these measures suffice for the data layer, too? I’d say they don’t.</p><p>While the data infrastructure can be secured using the tools mentioned above, data itself needs several other layers of protection within the organization to prevent, among other things, unauthorized access and accidental sharing of data. To do that, data teams usually had to rely on an external data governance tool before some data warehouse and data lake platform companies started offering an integrated data governance solution, Snowflake being one of the prominent ones.</p><p>Snowflake is unique in how it approaches data governance — the same way it approaches infrastructure — everything is done using SQL. The basic knowledge of SQL is something you can, more or less, take for granted in teams that directly deal with data, which is why this makes using Snowflake a desirable proposition for businesses. Snowflake’s governance features are not a replacement for a full-fledged governance tool, but it does some of the heavy lifting in certain areas. Snowflake’s governance features help you answer the following questions and more:</p><ul><li>Which business categories do your data objects belong to? This is done using <em>object tagging</em>.</li><li>Which columns contain sensitive, especially PII and PHI data? Snowflake’s built-in <em>data classification</em> engine can help you do that.</li><li>Which data objects should be visible to a user? Granular control of data objects is enabled by applying <em>row access policies</em> and <em>column-level security</em> features.</li><li>Which users are reading data from and writing data? Snowflake logs all such activity in the ACCESS_HISTORY table.</li><li>How has a data asset (table or column) transformed in a data pipeline after cleansing, wrangling, and transformation? Both ACCESS_HISTORY and OBJECT_DEPENDENCIES give you a detailed answer to that question.</li></ul><p>This article will introduce you to the data above governance themes with examples to help you navigate your data governance journey on Snowflake. Let’s dive right in!</p><h3>Tagging objects in Snowflake</h3><p>Tags are general-purpose labeling constructs and provide a clean and highly flexible solution for categorizing data assets, from coarse-grained organization-level objects like accounts and warehouses to fine-grained schema-level objects like tables, views, and columns. Tags can be used in several ways, some prescribed by Snowflake as listed below, while others are left for you to discover for your business use case:</p><ul><li>Track cost and usage of org-wide data assets</li><li>Classify and group data assets based on custom requirements</li><li>Apply data masking policies on columns with certain tags</li><li>Protect sensitive data using system-defined tags</li></ul><p>In addition to governance, tags help with many other use cases, such as search and discovery in data catalogs, business intelligence tools, etc. Many of these tools, such as <a href="https://docs.selectstar.com/integrations/snowflake/snowflake-tag-sync">Select Star</a>, <a href="https://docs.thoughtspot.com/software/latest/dataflow-snowflake-sync">Immuta</a>, and <a href="https://productresources.collibra.com/docs/collibra/latest//Content/Catalog/RegisterDataSource/Edge/SynchronizeSchemas/co_about-synchronizing-schemas.htm">Collibra</a>, have a two-way tag sync functionality, where you can apply tags from your data catalogs and they’ll get updated in Snowflake. This deserves mention because, in most cases, data catalogs only extract metadata from data sources and don’t write back to them. Think of this as out-of-the-box reverse ETL for metadata, i.e., the data catalog enriching the data source back.</p><h3>Identifying sensitive data</h3><p>While tags allow you to arbitrarily create labels for your data assets, in many cases, you would also need to classify your data assets based on their content. To do that, you can use Snowflake’s data classification feature. The classification process <a href="https://docs.snowflake.com/en/user-guide/governance-classify-concepts#compute-costs">requires a certain amount of compute power</a>, so you’ll need to have a warehouse up and running to use this feature. You can use this feature to:</p><ul><li>Identify sensitive data using native Snowflake features</li><li>Have oversight of sensitive data</li><li>Invoke external cloud-based functions for classification</li></ul><p>The core step of the process is extracting semantic categories that represent personal attributes, such as name, address, age, salary, and so on. To do that, you can call the <a href="https://docs.snowflake.com/en/sql-reference/functions/extract_semantic_categories">EXTRACT_SEMANTIC_CATEGORIES</a> function and, using a sample of the column data, and it will give you the probability of each column containing sensitive data. Aside from that, it provides you attributes like confidence, coverage, and <a href="https://docs.snowflake.com/en/user-guide/governance-classify-concepts#label-classification-system-tags">two system-defined tags</a>, SEMANTIC_CATEGORY and PRIVACY_CATEGORY.</p><p>Data classification isn’t a fully automatic process; it goes through a cycle of analysis, review, and application, where Snowflake takes care of steps one and three. However, it falls upon the data engineers to review whether Snowflake’s correct interpretation of the column data. After the review, you can use the <a href="https://docs.snowflake.com/en/sql-reference/stored-procedures/associate_semantic_category_tags">ASSOCIATE_SEMANTIC_CATEGORY_TAGS</a> stored procedure with the name of the table and the output of the EXTRACT_SEMANTIC_CATEGORIES function as parameters to automatically associate the tags to the table columns. You can use these tags to protect sensitive data using Snowflake’s data masking capabilities.</p><h3>Granular access to data</h3><p>While <a href="https://docs.snowflake.com/en/user-guide/security-access-control-overview#roles">RBAC (role-based access control)</a> and DAC (discretionary access control) allow you to manage object-level access in Snowflake, there are two other methods that you can use to restrict access to data in tabular objects, such as tables, views, and materialized views. Some examples of where granular access to data would make sense are when you want to:</p><ul><li>Restrict access to sales records for one region by members of other regions</li><li>Create an information barrier between teams to comply with regulatory requirements</li><li>Protect PII and PHI data by masking fields partly or in full based on various factors</li></ul><p>With row access policies and masking policies, you can restrict access to rows and columns. Both these methods allow for on-the-fly evaluation of policies. Based on these evaluations, you’ll be granted access to specific rows or columns. Row access policies use both <a href="https://docs.snowflake.com/en/sql-reference/expressions-conditional">Conditional Expression functions</a> and <a href="https://docs.snowflake.com/en/sql-reference/functions-context">Context Functions</a> for policy enforcement. For instance, you can check for the role a user is using to access the data by using the CURRENT_ROLE() context function.</p><h3>Mapping data lineage</h3><p>While implementing the modern data stack, it’s usual for a data cataloging or lineage tool to be set up to allow business and technical users to understand the flow of data and how it’s been transformed from source to target. Snowflake’s data lineage metadata comes in handy when you have to:</p><ul><li>Visualize how the data loads from sources, integrates with other sources, and becomes usable for the business</li><li>Perform impact analysis when making changes to data objects</li><li>Debug ETL issues that end up in stale, erroneous, and incomplete data</li></ul><p>Snowflake’s data lineage capabilities rest upon two distinct features; the metadata for both is captured in the <a href="https://docs.snowflake.com/en/sql-reference/account-usage">ACCOUNT_USAGE</a> schema. The <a href="https://docs.snowflake.com/en/sql-reference/account-usage/object_dependencies">OBJECT_DEPENDENCIES</a> view allows you to look at the dependency graph between objects, i.e., it shows you which objects need which other objects for proper functioning. This is helpful when planning to handle the changes to an upstream object and the impact those changes would have on the downstream objects.</p><p>The other feature is where Snowflake logs a year’s worth of query history in the <a href="https://docs.snowflake.com/en/sql-reference/account-usage/access_history">ACCESS_HISTORY</a> view. This view provides details about the query, such as which tables, views, materialized views, and columns were accessed. It also provides the text of the query in full. Using access history, you can also use external tools with their native SQL parsers to build or enrich data lineage as and when required.</p><h3>Conclusion</h3><p>This article walked you through Snowflake’s native data governance features. It also talked about identifying and protecting sensitive data, restricting access to data using various control measures, and using the metadata to build the lineage graph. For more, check out <a href="https://docs.snowflake.com/en/guides-overview-govern">Snowflake’s guide to data governance</a>, which talks about these capabilities in much more detail. Also, check out the latest <a href="https://medium.com/snowflake/tagged/data-governance">data governance-related blog posts</a> in the community-driven Snowflake blog on Medium.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9cdcd9468f5d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/snowflake-data-governance-overview-9cdcd9468f5d">Snowflake Data Governance Overview</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why less can be so much more when training a machine learning model]]></title>
            <link>https://medium.com/versent-tech-blog/why-less-can-be-so-much-more-when-training-a-machine-learning-model-1ab384758198?source=rss-36669c51ff0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ab384758198</guid>
            <category><![CDATA[machine-learning-ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Versent]]></dc:creator>
            <pubDate>Tue, 12 Sep 2023 20:13:46 GMT</pubDate>
            <atom:updated>2023-09-20T05:48:12.900Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Jayden Kur (jayden.kur@versent.com.au)</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GvvE9vQ_j7nTfcUU-z88xg.jpeg" /></figure><p><em>Training a machine learning (ML) model on poor quality data will inevitably produce poor quality outputs. This is commonly known as garbage in garbage out. Often developers create cutting-edge, best-practice machine learning models with access to all the compute power they can dream of, yet are provided with terrible low quality training data in abundance. They are then questioned on the garbage output. Being supplied a greater quantity of poor quality data is not the solution here and will not guarantee favourable outcomes.</em></p><h4>Focus on quality not quantity</h4><p>There is a misconception that more data will be better for training models, instead of focusing on the quality of the data. I have seen this repeatedly in industry, especially within enterprise level clients such as production, energy and education. All these industries are, to some degree, are guilty of supplying poor quality data to engineers and expecting magic. This was even the case when I worked for a leader in the AI industry that had implemented a ML solution in one of Australia’s largest mining companies.</p><p>I have also experienced this in my own education where I was using convolutional neural networks (CNN) for identifying morphological properties of galaxies. I was given images so blurry that I could not tell if it was a picture of a galaxy, or someone had gone out at night and taken photos of a distant streetlamp on a flip phone. I quickly realised if I wanted to produce anything useful, I would need to secure higher quality training data. I reached out to the owner of a large collection of labelled galactic images in America. I secured a much higher quality and resolution data set of labelled galactic images to train my model. I was able to achieve an accuracy of close to 100% using drastically less images for training model. It has since been used to classify millions of morphological properties of galaxies which are usable by physicists.</p><p>I recently participated in our annual company hackathon, where my team created an educator bot with an initial lens of onboarding. We leveraged Generative AI models and intelligent data retrieval to offer a fresh and novel approach to employees accessing, processing and understanding information. Our first optimistic approach was to give it access to all non-critical documents, ask it questions and fact checked the response. This approach did not work and was, well, garbage. We were then selective about the documents provided and restricted the questions that could be asked of it. This proved to be dramatically better than our initial approach. Overall, we found that supplying less data of higher quality produced more useful and accurate responses to our users.</p><h4>Key learnings: a smaller quantity of higher quality data is actually more</h4><p>As we leverage increasing ML technologies in our lives, the key learning from my work and research is that when training a machine learning model, a smaller quantity of higher quality data is actually more.</p><blockquote>We must be intentional. We cannot throw garbage at these models and expect miracles.</blockquote><p>Engineers in industry do not always have the luxury to secured higher quality data and often it does not exist at all. Industry leaders should take it upon themselves to take action to secure the highest quality data for optimal outcomes.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ab384758198" width="1" height="1" alt=""><hr><p><a href="https://medium.com/versent-tech-blog/why-less-can-be-so-much-more-when-training-a-machine-learning-model-1ab384758198">Why less can be so much more when training a machine learning model</a> was originally published in <a href="https://medium.com/versent-tech-blog">Versent Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>