<?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 Lito Kusnadi on Medium]]></title>
        <description><![CDATA[Stories by Lito Kusnadi on Medium]]></description>
        <link>https://medium.com/@lito-kusnadi?source=rss-cb0bd1ede671------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*BCTzhOR1l6cDZgMHWQO2dQ.jpeg</url>
            <title>Stories by Lito Kusnadi on Medium</title>
            <link>https://medium.com/@lito-kusnadi?source=rss-cb0bd1ede671------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 17:38:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lito-kusnadi/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[Google + Auth0 as External Identity Provider for AWS SSO]]></title>
            <link>https://medium.com/carsguide-engineering/google-auth0-as-external-identity-provider-for-aws-sso-61d2e32ee6a4?source=rss-cb0bd1ede671------2</link>
            <guid isPermaLink="false">https://medium.com/p/61d2e32ee6a4</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[auth0]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[single-sign-on]]></category>
            <category><![CDATA[aml]]></category>
            <dc:creator><![CDATA[Lito Kusnadi]]></dc:creator>
            <pubDate>Tue, 03 Nov 2020 03:52:14 GMT</pubDate>
            <atom:updated>2020-11-03T04:30:16.010Z</atom:updated>
            <content:encoded><![CDATA[<p>I was working under a very unique scenario to come up with this solution and hopefully someone, with a similar situation, can benefit from this guide.</p><p>I have to do a proof of concept to manage the developer’s access to our AWS accounts. In this scenario, there is already a working Single Sign-On (SSO) system with Multi-Factor-Authentication (MFA) against a user directory such as Active Directory. However, my team has no administrative access to manage the existing Active Directory. The business emails that are controlled through the user directory can be used with Google products (such as Google Drive).</p><p>So here is the question:</p><blockquote>How to manage user access to the AWS accounts using a centralised authentication system without administrative privilege to the user directory?</blockquote><p>The answer is simple: leverage social media login. Since the business email can be used to authenticate using Google, we can use Google OAuth to grant access to AWS SSO.</p><p>Before we continue, some of you may think: letting social media access to the AWS accounts is a <strong>very bad</strong> idea.</p><p>I completely agree if we do not build any security around it. Social media access provides great flexibility for authentication; this is the one feature that we want to take advantage from. I will show you ways to vet social media users so that only trusted users can access your AWS accounts.</p><p>This solution does not come without its drawback. I will explain later for you to judge if this whole thing worths the effort.</p><h3>Architecture</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_WxkL0V81grHePYw1LRFUQ.png" /><figcaption>Google + Auth0 + AWS SSO Integration</figcaption></figure><p>Because the AWS accounts are managed in a hierarchal fashion, we can use AWS SSO product to centrally manage access to the other linked AWS accounts. Let’s go through the architectural diagram:</p><ol><li>Users go to User Portal URL provided by AWS SSO.</li><li>AWS redirects users to an Auth0 login form (this is essentially an Auth0 application).</li><li>The users validate their credentials using Google.</li><li>Google verifies the access and create the user record in Auth0 Database.</li><li>Auth0 redirects to an AWS SSO portal where the users can see the available accounts permitted to them.</li><li>When the users choose an account to log into, AWS SSO checks the mapping between users, accounts, and permission sets in AWS SSO.</li><li>The mapping between users, accounts, and permission sets must be created manually. This is our main security layer. If the users have no mapping, then the redirection will fail in step 5.</li><li>When the mapping between users and accounts is created, AWS SSO creates an IAM Identity Provider in the target linked AWS account. This lets AWS SSO to log into the target account using SAML.</li><li>When the mapping between users and permission sets is created, AWS SSO creates an IAM Role in the target linked AWS account. This is the role that a user will assume when logging into the linked AWS account.</li></ol><p>To improve the security, we can use Auth0 Rule to whitelist email domain. Any random email authenticated through social media will be blocked at Auth0.</p><p>Remember the drawback I mentioned earlier? Here it is: in step 4 above, Auth0 creates a user record in Auth0 Database when the user is validated via social media, even if there’s a whitelist rule. So your Auth0 Database can potentially be littered with spammer emails. The solution: use AWS Lambda to clean up the Auth0 Database on a daily basis via Auth0 API. The AWS Lambda clean up part is not covered in this guide.</p><p>Now, let’s do the fun bit. How can you configure this contraption?</p><p>Be aware that you will be jumping back and forth between AWS, GCP, and Auth0 during configuration. So strap yourself!</p><h3>AWS SSO Configuration</h3><p>Go to the main (paying) account in AWS and enable the AWS SSO (if this is a production account, I would suggest you to understand what AWS SSO will do to your account).</p><p>Go to <strong>AWS SSO</strong> &gt; <strong>Settings</strong> and change the <strong>Identity Source</strong> to <strong>External Identity Provider</strong>.</p><p>Show the <strong>Individual Metadata Values</strong> and note down the <em>Sign-in URL, ACS URL, and Issuer URL</em>; these will be required in Auth0.</p><p>Keep this section open and let’s jump to Auth0.</p><h3>Auth0 New Application</h3><p>Choose an Auth0 tenancy or create a new tenancy. Personally I would prefer using a new tenancy.</p><p>Create a new application with <strong>Regular Web Application</strong> as its type; you can name it anything you want.</p><p>Put the <strong>ACS URL</strong> from AWS SSO as Auth0’s <strong>Allowed Callback URLs</strong>.</p><p>Click <strong>Save</strong>.</p><p>Go to <strong>Add-on</strong> tab and enable <strong>SAML2 Web App</strong>.</p><p>Use the following JSON in the setting of SAML2 web app (don’t forget to replace the <strong>ACS URL</strong> and <strong>Issuer URL</strong>):</p><pre>{<br>  &quot;audience&quot;: &quot;<strong>AWS_SSO_ISSUER_URL</strong>&quot;,<br>  &quot;destination&quot;: &quot;<strong>AWS_SSO_ACS_URL</strong>&quot;,<br>  &quot;mappings&quot;: {<br>    &quot;email&quot;: &quot;<a href="http://schemas.xmlsoap.org/ws/2005/05/identity/claims/emailaddress">http://schemas.xmlsoap.org/ws/2005/05/identity/claims/emailaddress</a>&quot;,<br>    &quot;name&quot;: &quot;<a href="http://schemas.xmlsoap.org/ws/2005/05/identity/claims/name">http://schemas.xmlsoap.org/ws/2005/05/identity/claims/name</a>&quot;<br>  },<br>  &quot;createUpnClaim&quot;: false,<br>  &quot;passthroughClaimsWithNoMapping&quot;: false,<br>  &quot;mapUnknownClaimsAsIs&quot;: false,<br>  &quot;mapIdentities&quot;: false,<br>  &quot;nameIdentifierFormat&quot;: &quot;urn:oasis:names:tc:SAML:1.1:nameid-format:emailAddress&quot;,<br>  &quot;nameIdentifierProbes&quot;: [<br>    &quot;<a href="http://schemas.xmlsoap.org/ws/2005/05/identity/claims/emailaddress">http://schemas.xmlsoap.org/ws/2005/05/identity/claims/emailaddress</a>&quot;<br>  ]<br>}</pre><p>This above JSON can be found <a href="https://auth0.com/docs/integrations/aws/configure-amazon-web-services-for-sso">here</a>.</p><p>Scroll down and click <strong>Enable</strong>.</p><p>Go to the <strong>Usage</strong> tab and download the <strong>Identity Provider Metadata</strong> XML. This file will be used by AWS SSO.</p><p>Close the Add-on page.</p><p>Back to AWS SSO.</p><h3>AWS SSO Configuration (continue…)</h3><p>Click <strong>Browse</strong> for <strong>Identity Provider Metadata</strong> and upload the XML file from Auth0.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*39slnxkwezXqUK8s0i1yIg.png" /><figcaption>AWS SSO IdP SAML Metadata</figcaption></figure><p>Click <strong>Next: Review</strong> and type <strong>ACCEPT</strong> to confirm the change of identity source.</p><p>After returning to the AWS SSO setting page, click <strong>Cancel</strong> to get out of the SSO identity source page (there’s room for improvement here, AWS).</p><h3>Google Cloud Platform OAuth Credential</h3><p>Auth0 uses its own dev OAuth key to authenticate with Google. You need to use your own OAuth key to make this configuration works. Let’s create an OAuth credential key.</p><p>Create a new project in GCP. A project can only have one OAuth consent screen. This must be set to Auth0 domain.</p><p>Go to <strong>API and Services</strong> &gt; <strong>Credentials</strong> in GCP.</p><p>Click <strong>Create Credential</strong> and select <strong>OAuth Client ID</strong>.</p><p>Click <strong>Configure Consent Screen</strong> and provide the following values:</p><ul><li>Application Type : Public</li><li>Application Name : auth0_aws_sso (anything unique)</li><li>Authorized Domains : auth0.com</li></ul><p>Click <strong>Save</strong>.</p><p>Continue with <strong>OAuth Client ID</strong> creation and set the following values:</p><ul><li>Name : auth0 aws sso (anything unique)</li><li>URIs : https://YOUR_DOMAIN (this is from Auth0 Application)</li><li>Authorized Redirect URIs : https://YOUR_DOMAIN/login/callback</li></ul><p>Note down the <strong>Client ID</strong> and <strong>Client Secret</strong> of the new key.</p><p>These steps can be found <a href="https://auth0.com/docs/connections/social/google">here</a>.</p><p>Back to Auth0.</p><h3>Completing Auth0 Settings</h3><p>In Auth0, go to <strong>Connection</strong> &gt; <strong>Social</strong> and select <strong>google-oauth2</strong>.</p><p>Provide the <strong>Client ID</strong> and <strong>Client Secret</strong> of the OAuth Client ID from GCP.</p><p>Click <strong>Save</strong>.</p><p>Go to <strong>Rules</strong> and create a new rule (there’s a template for email domain whitelist).</p><p>You can use this <a href="https://auth0.com/rules/simple-domain-whitelist">code</a> as an example.</p><p>Congratulation! Auth0 setting is complete.</p><p>Now, we need to create the users and permission sets in AWS SSO. This is a manual process to provision trusted users to your AWS accounts.</p><h3>AWS SSO User and Permission Set</h3><p>Go back to AWS SSO.</p><p>Go to <strong>AWS SSO</strong> &gt; <strong>AWS Accounts</strong> &gt; <strong>Permission Sets</strong>.</p><p>Follow the instruction on the screen to create the permission set.</p><p>Go to <strong>AWS SSO</strong> &gt; <strong>Users</strong> and click <strong>Add User</strong>.</p><p>Use the <strong>email address</strong> in the following fields:</p><ul><li>Username</li><li>Email Address</li></ul><p>Skip group assignment and add the user.</p><p>Go to <strong>AWS SSO</strong> &gt; <strong>AWS Accounts</strong> and choose the <strong>AWS Organization</strong> tab.</p><p>Select an account and assign the user with a permission set.</p><h3>Test User Login</h3><p>Use the <strong>User Portal URL</strong> to get a login screen.</p><p>An Auth0 application login is shown and click <strong>Sign in with Google</strong>.</p><blockquote>If the existing authentication already supports MFA, it will ask for MFA verification.</blockquote><p>Users will see a portal where they can choose the account to log into, along with the allocated permission set.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RXCsmhI6GXL4n1DLbMCLQA.png" /></figure><p>Note that there is the <strong>command line or programmatic access</strong> option. This means you can use this for AWS CLI as well using the AWS STS token given to the existing session.</p><p>You can either get the token from the portal and edit the ~/.aws/credentials manually or you simply use the following command:</p><pre>aws configure sso<br>SSO start URL [None]: **provide user portal URL**</pre><h3>How to Troubleshoot</h3><p>You can look at the logs in Auth0 web console to understand what is going on.</p><p>There are two common errors that I have seen so far:</p><ul><li>MFA is invalid : This is because the user has been verified by social media but no matching user can be found in AWS SSO. This is a good thing from security point of view.</li><li>Invalid input : Auth0 Google connector is using Auth0 OAuth dev key; change the key to your own OAuth key.</li></ul><p>Hope this guide can give more ideas on how to authenticate users to AWS accounts.</p><p>Have fun and be safe.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61d2e32ee6a4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/carsguide-engineering/google-auth0-as-external-identity-provider-for-aws-sso-61d2e32ee6a4">Google + Auth0 as External Identity Provider for AWS SSO</a> was originally published in <a href="https://medium.com/carsguide-engineering">CAMS Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introduction to Sceptre: An AWS Cloudformation Orchestration Tool]]></title>
            <link>https://medium.com/carsguide-engineering/introduction-to-sceptre-an-aws-cloudformation-orchestration-tool-4b8453c0ae81?source=rss-cb0bd1ede671------2</link>
            <guid isPermaLink="false">https://medium.com/p/4b8453c0ae81</guid>
            <category><![CDATA[respect]]></category>
            <category><![CDATA[orchestration]]></category>
            <category><![CDATA[cloudformation]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[engineering]]></category>
            <dc:creator><![CDATA[Lito Kusnadi]]></dc:creator>
            <pubDate>Wed, 13 Jun 2018 21:06:11 GMT</pubDate>
            <atom:updated>2018-06-21T02:24:40.767Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PUEaV2CxRP3cZlGz" /><figcaption>Photo by <a href="https://unsplash.com/@dmpop?utm_source=medium&amp;utm_medium=referral">Dmitri Popov</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The term “infrastructure as code” has been pervasive in the development and operation space for quite some time. It has been a never ending quest to find the best breed of tools to materialize infrastructure as code in the most simplistic way.</p><p>Cloudformation is an AWS service that allows to spin up any resource in AWS with a predefined set of blueprints. By using CloudFormation, the recipe of resource creation is self-documenting and changes to the resources can be tracked by the help of code repository.</p><h3>Why Sceptre?</h3><p>Cloudformation is not pretty; it is composed using JSON or YAML. It can become so big and convoluted to debug. And very often there are variables that must be injected dynamically during runtime into Cloudformation which can be out of the scope of Cloudformation’s ability. Sceptre is a tool to abstract Cloudformation creation and it comes with a lot of features to make Cloudformation orchestration becoming more powerful.</p><p>Unlike Terraform (which is also another Cloudformation abstraction tool), Sceptre does not track state. This implies that what is written in Sceptre is an additive-write to AWS service. For example, if a user defines resource A and B with Sceptre and runs it against Cloudformation service, AWS will create resource A and B respectively. However, when a user changes the definition to create resource B and C, the resulting resources in AWS will be resource A, B and C. Resource A still prevails in AWS.</p><p>With Terraform, which tracks states, if the above scenario is applied, the resulting resources will be only resource B and C in AWS. So why is Sceptre preferred over Terraform? To have a state, users must also maintain the state. In a multi user environment, maintaining a state can be an overhead. When a state is corrupted, the Cloudformation abstraction code cannot be run against the same resource. This is a big problem. Therefore, backup must be performed against the state regularly. The nice thing about Sceptre is simplicity: it is a tool to abstract Cloudformation creation, nothing more. There is no state to refer to; thus Sceptre users will always be able to run their codes against existing infrastructure.</p><p>Moreover, slicing a big Cloudformation template into manageable pieces can be done with ease in Sceptre. When additional required jobs are not covered in Cloudformation, users can easily develop their own custom codes around Sceptre through Hooks and Resolvers.</p><h3>Structure of Sceptre</h3><p>The simplest form of Sceptre structure is 2 directories:</p><ul><li>config: this is environment specific-template configuration which usually holds various values in Cloudformation parameters</li><li>templates: this is the location of reusable Cloudformation templates that can span across multiple environments</li></ul><p>An example of directory structure in Sceptre:</p><blockquote>config/prod/billing-api.yaml<br>config/dev/billing-api.yaml<br>templates/billing-api-main.yaml</blockquote><p>The main template (in this case, billing-api-main.yaml) is simply a standard CloudFormation file either in JSON and YAML format. The config environment file (such as config/prod/billing-api.yaml) overwrites the parameters or resources, with a reference to the main template file. On runtime, these two files will be merged and become a single Cloudformation blueprint.</p><p>config/prod/billing-api.yaml:</p><blockquote>template_path: templates/billing-api-main.yaml<br>parameters:<br> keypair: prod-key.ssh<br> instancesize: m4.large</blockquote><p>Optionally, in Sceptre, users can extend the functionality of Cloudformation that cannot be done natively through Hooks and Resolvers. Both Hooks and Resolvers are written using Python. Hooks are custom codes that are executed based on events such as after creation, before update, after deletion, etc. While Resolvers are essentially functions that feed input to Cloudformation. To use Hooks and Resolvers, users just simply create “hooks” and “resolvers” directory respectively and put their codes inside them.</p><p>To run Sceptre on local machines, developers can simply:</p><ul><li>Install Anaconda Python</li><li>Install Sceptre via pip</li></ul><p>Example of environment preparation by using conda package manager from Anaconda Python:</p><blockquote>conda create -n mysceptre python=3.6.3 ipython<br>(Mac) source activate mysceptre<br>(Windows)activate mysceptre<br>pip install sceptre<br>sceptre -h</blockquote><p>Give Sceptre a go and see what you can build.</p><p>Links:</p><ul><li><a href="https://anaconda.org/anaconda/python">https://anaconda.org/anaconda/python</a></li><li><a href="https://github.com/cloudreach/sceptre">https://github.com/cloudreach/sceptre</a></li><li><a href="https://sceptre.cloudreach.com/latest/docs/get_started.html">https://sceptre.cloudreach.com/latest/docs/get_started.html</a></li><li><a href="https://sceptre.cloudreach.com/latest/docs/architecture.html">https://sceptre.cloudreach.com/latest/docs/architecture.html</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4b8453c0ae81" width="1" height="1" alt=""><hr><p><a href="https://medium.com/carsguide-engineering/introduction-to-sceptre-an-aws-cloudformation-orchestration-tool-4b8453c0ae81">Introduction to Sceptre: An AWS Cloudformation Orchestration Tool</a> was originally published in <a href="https://medium.com/carsguide-engineering">CAMS Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[RESTful API Development with SLA]]></title>
            <link>https://medium.com/carsguide-engineering/restful-api-development-with-sla-452fdf58d5cd?source=rss-cb0bd1ede671------2</link>
            <guid isPermaLink="false">https://medium.com/p/452fdf58d5cd</guid>
            <category><![CDATA[stress-testing]]></category>
            <category><![CDATA[load-testing]]></category>
            <category><![CDATA[siege]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Lito Kusnadi]]></dc:creator>
            <pubDate>Tue, 21 Nov 2017 00:24:29 GMT</pubDate>
            <atom:updated>2017-11-21T00:25:05.666Z</atom:updated>
            <content:encoded><![CDATA[<p>When developing RESTful API endpoint, programmers want to focus on business logic. The performance of the API endpoint usually comes as part of the solution at the later stage during the performance testing. This, however, creates a slower development cycle as testers need to get back to the developers for optimization requests.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/580/1*osihPt0-oQoLjxOUElcqkQ.png" /></figure><p>Preparing performance testing is time consuming; but with the right tool, performance can be part of the quality of the API endpoints being developed without the time overhead. We focus on one stress tool called <strong>siege</strong>. This is a simple command line that can be used at the developer’s end to ensure the API endpoint meets the response requirement as part of the endpoint SLA.</p><p>It is important to note that developers should not try to do stress testing with siege, they should constantly measure the API endpoint response time during development.</p><p>Siege is lightweight and very simple to run. The application can be obtained directly from <a href="https://github.com/JoeDog/siege">GitHub</a> and compiled from source code. This may involve compiler installation and other development libraries if you are running on Linux. Developers who use OSX can install siege binary by using <em>homebrew</em>. The command is:</p><blockquote>/&gt; brew install siege</blockquote><p>For developers on Windows, you may switch to Linux or OSX. :)</p><p>Developers can use siege to benchmark the API endpoint. The command is:</p><blockquote>/&gt; siege -b -c5 <a href="http://someurl.com/path/page">http://someurl.com/path/page</a></blockquote><blockquote>/&gt; siege -d1 -c5 <a href="http://someurl.com/path/page">http://someurl.com/path/page</a></blockquote><p>Both commands above are equivalent. Developers can stop it with sending CTRL+C keys. This will perform a benchmark of 5 concurrent users with delay of 1 second (to concentrate requests within the shortest delay possible). If there is requirement to pass header value or anything esoteric, please consult with the siege manual.</p><p>The three important metrics to check from the result are:</p><p>Availability</p><p>This shows if all the requests are successful. If not, check the infrastructure (such as EC2 or network connectivity), or the endpoint itself (application uses too much resources). The target is to achieve 100% availability.</p><p>Response time</p><p>This shows how quick the API endpoint is. This is the SLA developers must keep in mind. Set a target, say, 300ms and try to achieve it.</p><p>Transaction rate</p><p>This shows the throughput of the API endpoint. The higher the throughput, the more efficient the API endpoint is. This is because the system finishes faster to provide a response and is able to move on to the next request.</p><p>Now developers can frequently use the tool to check their API performance with ease during development; thus, making performance quality inherent to the process.</p><h3>Creating Load Test with Siege</h3><p>For testers who want to find out how the application perform in production scenario, use this command:</p><blockquote>/&gt; siege -d10 -c[concurr_num] -t2M -f urls.txt -i</blockquote><p>Parameters explanation:</p><ul><li>-d10: Use 10 seconds delay to the concurrent number of connections so they will not go in at once; this gives more realistic scenario of day to day usage</li><li>-cX: Start with small concurrent number and confirm they are working without issue, then increment it slowly until the system response breaks</li><li>-t2M: Sustain the load test for 2 minutes for easy visibility</li><li>-f urls.txt: input urls from a text file</li><li>-i: internet mode introduces randomness to the requests</li></ul><p>Tips:</p><ul><li>When load testing with EC2 instances, avoid t2 instance type because these instances have CPU credits that may skew the test result</li><li>When running siege from an EC2 instance to gain superb network connectivity, make sure the siege runner has bigger instance type than the target</li><li>Use NewRelic agent when doing load testing to make debugging easier</li></ul><p>What to Avoid:</p><p>When load testing an application, avoid testing until the breaking point. When the test have 500 responses from the application, this is sufficient to tell the limit of the application. The benchmark should take lower concurrency limit to the point that the application returns 200 responses. Should the benchmark be using higher concurrency limit, the test is simply measuring the limit of OS sockets (which is predefined in the OS).</p><p>The main reason OS runs out of sockets is that the application is too slow (in respect to the request rate), that it does not release the socket for the next requests in a timely manner.</p><p>Remember, the aim is to find out how much the server can take <em>while</em> <em>fulfilling</em> the SLA requirements.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=452fdf58d5cd" width="1" height="1" alt=""><hr><p><a href="https://medium.com/carsguide-engineering/restful-api-development-with-sla-452fdf58d5cd">RESTful API Development with SLA</a> was originally published in <a href="https://medium.com/carsguide-engineering">CAMS Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>