<?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 Guy Rombaut on Medium]]></title>
        <description><![CDATA[Stories by Guy Rombaut on Medium]]></description>
        <link>https://medium.com/@guyromb?source=rss-4ae0aed9dc32------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*9E7f6d4V_hJ8Un_TY4hRCw.jpeg</url>
            <title>Stories by Guy Rombaut on Medium</title>
            <link>https://medium.com/@guyromb?source=rss-4ae0aed9dc32------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 06:29:27 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@guyromb/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[Big Data and AI as decision making framework, and few related notes on the Coronavirus]]></title>
            <link>https://medium.com/@guyromb/big-data-and-ai-as-decision-making-framework-and-few-related-notes-on-the-coronavirus-8f0f2aa6a2ab?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f0f2aa6a2ab</guid>
            <category><![CDATA[decision-making]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Thu, 20 Feb 2020 09:37:33 GMT</pubDate>
            <atom:updated>2020-02-20T09:37:33.550Z</atom:updated>
            <content:encoded><![CDATA[<p>As a manager, data helps me decide on the basis of evidence rather than intuition. Data revolutionise strategic thinking.</p><p>Velocity of live data is critical to be agile (make decision before competitor does), variety of sources helps you see the full picture etc.</p><p>At OneFit, we are streaming data from more than 50 different sources (internal and external) using Kafka. We store it in one place (Google Big Query), and analyse it for different purposes (Using Kubeflow ML pipelines, and Looker for modelling and visualisation).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/603/0*IKjmlFIvf9-4dRzh" /></figure><p>This is a simple yet powerful concept. It is known that intelligence arises from connecting the dots. At OneFit, the business intelligence arises by connecting different kinds of data. By joining a plethora of different datasets and analysing it, we extract deep insights that were never visible before. We use a large spectrum of analytics and machine learning techniques to enable the entire company for data driven decisions. In every step of any project, data insights are our driving power; from smart marketing strategies to product optimisations, from smart workload allocation strategies to financial forecasting, and so on. Some concrete examples that big data helps us are measuring performance, validate our product and pricing strategies, predict available spots for inventory, improving customer retention, enhance our user experience with recommendations and personalised banners and much more.</p><p>This is possible thanks to great cooperation between cross-functional teams — Tech team, Product team and together with our Head of BI &amp; Data, Dr. Ben E. Kuzey and his team who enable our data-driven culture. Together, we <em>Research &amp; Develop</em>.</p><p>Sometimes, surprisingly, there are still skepticism around the topic of AI &amp; Data. Google and Amazon could be a great example for very successful companies who leverage big data technologies. But they could not be used as an example for all: industries are different, companies’ business models varies and so on.</p><h3>“Show me the evidence that using AI will improve business performance”</h3><p>It’s hard to answer this question rigorously, which is slightly embarrassing. Luckily, MIT, Harvard and McKinsey’s already made various research around this topic, as part of their strategic advise to companies.</p><p>Hypothesis:<em> Would data-driven companies perform better?</em></p><p>They conducted structured interviews with executives at 330 public North American companies about their organisational and technology management practices, and gathered performance data from their annual reports and independent sources [5].</p><blockquote>Not everyone was embracing data-driven decision making. In fact, they found a broad spectrum of attitudes and approaches in every industry. But across all the analyses one relationship stood out: The more companies characterised themselves as data-driven, the better they performed on objective measures of financial and operational results.</blockquote><p>In particular, companies in the top third of their industry in the use of data-driven decision making were, on average, 5% more productive and 6% more profitable than their competitors [3].</p><h3>The Coronavirus</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FA9FnELPZ-0g_NDn" /></figure><p>Another interesting example of using data as a decision making framework, is the Coronavirus, which was spreading quickly in China, caused shutdowns of factories and even complete cities.</p><p>Investors are trying to estimate the impact on the economy. They are using unconventional data to measure the hit on the world’s second-largest economy’s, especially when the official data from the government can not be trusted (claimed to be covered up). They use data from food delivery apps transactions to air pollution levels (dropped by 36%), flights reservations (dropped by 70%) and number of drivers via TomTom or Waze [6]. When those parameters normalise, it could be an indication when to invest again. Random sources of data are used to draw a picture.</p><h3>Data-driven decisions are better decisions — it’s as simple as that.</h3><p>There is so much data, which humans cannot process themselves. AI is the engine, the intelligence, that helps you learn and leverage data insights better. It is a crucial tool that helps you make the right decision <em>fast</em>.</p><p>Nick Bostrom, in his book <em>Superintelligence: Paths, Dangers, Strategies</em>, introduced the paths and timelines when intelligent machines could improve their own capabilities faster than human computer scientists. His estimation based on multiple sources that this will happen within the next 5 years is likely, as presented in the following chart [5].</p><p>Businesses must be ready, the revolution already started.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*LwacY03a7K8gNoBa" /></figure><h3>Side Notes</h3><p>AI’s power does not erase the need for human vision. Strong leadership, talent and culture are essential to spot opportunities/threats and work together to strengthen the organisation’s weaknesses. Data/AI is just the tool or the framework to help you, similarly as other technologies. Intuition still plays an important role in understanding the context of data in order to prove a hypothesis (use it as evidence to base my decisions on). Without context, data is just a bunch of numbers. Intuition helps me shape the context.</p><p><em>Data and AI are the engine behind successful companies. I am proud to be part of such a company.</em></p><p><em>______</em></p><p><em>We are hiring data scientists, data engineers, and data analysts. Check our </em><a href="https://jobs.one.fit/o/data-analyst-1"><em>jobs website</em></a><em> for currently open positions.</em></p><p><em>______</em></p><p><em>Initially published in </em><a href="https://www.linkedin.com/pulse/big-data-ai-decision-making-framework-few-related-notes-guy-rombaut-/"><em>LinkedIn</em></a><em>.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ikabQ8YjnwtP__qomeSoEA.jpeg" /></figure><p>References:</p><p>[1] Strength in Numbers: How Does Data-Driven Decision-making Affect Firm Performance? <a href="http://ebusiness.mit.edu/research/papers/2011.12_Brynjolfsson_Hitt_Kim_Strength%20in%20Numbers_302.pdf">http://ebusiness.mit.edu/research/papers/2011.12_Brynjolfsson_Hitt_Kim_Strength%20in%20Numbers_302.pdf</a></p><p>[2] <a href="https://hbr.org/2020/02/data-driven-decisions-start-with-these-4-questions">https://hbr.org/2020/02/data-driven-decisions-start-with-these-4-questions</a></p><p>[3] <a href="https://hbr.org/2012/10/big-data-the-management-revolution">https://hbr.org/2012/10/big-data-the-management-revolution</a></p><p>[4] Nick Bostrom, in his book <em>Superintelligence: Paths, Dangers, Strategies</em></p><p>[5] Jeremy Howard, “<a href="https://www.ted.com/talks/jeremy_howard_the_wonderful_and_terrifying_implications_of_computers_that_can_learn?language=en">The wonderful and terrifying implications of computers that can learn</a>.”, TED.</p><p>[6] <a href="https://www.carbonbrief.org/analysis-coronavirus-has-temporarily-reduced-chinas-co2-emissions-by-a-quarter">https://www.carbonbrief.org/analysis-coronavirus-has-temporarily-reduced-chinas-co2-emissions-by-a-quarter</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f0f2aa6a2ab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Decode JWT Traefik access logs using Filebeat]]></title>
            <link>https://medium.com/@guyromb/decode-jwt-traefik-access-logs-using-filebeat-95d935eb7c4f?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/95d935eb7c4f</guid>
            <category><![CDATA[filebeat]]></category>
            <category><![CDATA[logs]]></category>
            <category><![CDATA[traefik]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Sat, 15 Feb 2020 10:19:38 GMT</pubDate>
            <atom:updated>2020-02-20T22:10:04.896Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Introduction and Use Case</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UsNCBgwD9-T2dzYsFZTxcQ.jpeg" /></figure><p>Sometimes it is useful to inspect API calls in order to debug a user related issue and reproduce end to end flow. This could be achieved by checking the server access logs.</p><p>Often authentication is done via JWT (JSON Web Tokens), added as a Bearer token in the headers. This token contains the user ID. Decoding the JWT in the logged requests, will help us get all requests made by a user.</p><p><strong>How JWT is structured?</strong></p><p>JWT contains three parts: header, payload and signature (h.p.s). Looking closely at the way the algorithm works shows that the first each of the two parts (h.p) are just base64 encoded, while the third part, the signature, is the only one which is actually encrypted (using a secret).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yfQ_qd5Ck998Z2FFYkXqZQ.png" /></figure><p><strong>Logs streaming</strong></p><p>We are using Logstash to stream logs to ElasticSearch. ElasticSearch allows us to find logs quickly from cross systems.</p><p>Logstash works on UDP. However, Traefik allows to send logs only to stdout or a file. Therefore, a log files streaming solution is required. We use Filebeat- a Lightweight shipper for logs from the family of Elastic stack.</p><p>Filebeat has many advantages. For example, it can be used in Kubernetes as a DaemonSet and listen to logs based on annotations. Also, it acts as a back-pressure regulator.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_IkDgzwoS5ATJL_FaOwmCw.png" /><figcaption>Source: <a href="https://www.elastic.co/beats/filebeat">https://www.elastic.co/beats/filebeat</a></figcaption></figure><p>Another main advantage of Filebeat is its processors, which allows to process the logs before they are shipped — enhance or filter. You can create custom scripts (using JavaScript), or use the <a href="https://www.elastic.co/guide/en/beats/filebeat/current/filtering-and-enhancing-data.html">pre-defined processors</a>.</p><p>We <a href="https://github.com/elastic/beats/pull/13144">introduced</a> decode_base64_field in Elastic 7.3.1. Among other use cases, this feature allows us to decode the JWT payload and extract the user ID.</p><p><strong>Traefik + Filebeat + JWT decoder</strong></p><p>TL;DR Let’s put everything together. If you already have Filebeat and Traefik installed. You can just add the following annotations to Traefik deployment:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8d01fdca0fbf68113ca0f6f8f594645c/href">https://medium.com/media/8d01fdca0fbf68113ca0f6f8f594645c/href</a></iframe><p>This will extract the user ID(sub) into a custom field (jwtPayloadJson.sub).</p><p>Now we can search all logs that are related to a user ID.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G8mjAVAuEt15uCaXoUytVA.png" /></figure><p>Make sure you didn’t forget to configure Traefik to send logs to stdout (via TOML). You can find more information about it <a href="https://docs.traefik.io/v1.7/configuration/logs/#access-logs">here</a>, or using the following basic example.</p><pre>[accessLog] <br>format = &quot;json&quot;</pre><p>Enjoy!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95d935eb7c4f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Security in the IoT generation: the Guy Rombaut point of view]]></title>
            <link>https://medium.com/@guyromb/security-in-the-iot-generation-the-guy-rombaut-point-of-view-c6f504e02253?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/c6f504e02253</guid>
            <category><![CDATA[internet-of-things]]></category>
            <category><![CDATA[security-by-design]]></category>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[backend]]></category>
            <category><![CDATA[security]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Mon, 21 Oct 2019 22:12:03 GMT</pubDate>
            <atom:updated>2019-10-21T22:20:31.904Z</atom:updated>
            <content:encoded><![CDATA[<h3>Security in the IoT generation: the OneFit CTO point of view</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*shpbTOS3L_JrPh6iI6Wy7A.jpeg" /><figcaption>Credit: <a href="https://medium.com/u/3fc6763cc122">Miriam Busch</a> (Karlmax Berlin)</figcaption></figure><p>The world is increasingly interconnected through the internet. More and more “smart” devices and appliances are now online 24 hours a day. But what are the real implications of the development of IoT and home automation? Are users really safe? What is the level of protection for their data? And what are the biggest risks to privacy?</p><p>During Codemotion Milan 2018, Guy Rombaut, CTO at OneFit, spoke to us in depth about these issues.</p><p>There are now many low-cost devices always connected and already installed in the workplace and home. Exposing these devices to the internet without precautions and without sensitising the users to their implications is a big risk. In addition to the possible violations of individual privacy, there is also a more relevant aspect concerning security.</p><p>In the absence of the necessary countermeasures, malicious users could take possession of our “smart” car, or turn off the electricity of a home remotely, or lock us out of the house. This is the potential downside of the domotics and IoT devices.</p><p>The well-known Ransomware WannaCry is a striking example of how cybersecurity is a topic too often ignored today. Although there is a lot of work being done in schools to make students aware of cybersecurity, very often the most basic security rules are not applied as much by individual users as by companies. So can a secure smart system be implemented? Obviously, yes. Just follow some basic rules that should be applied by all manufacturers on the market:</p><p>Always employ staff specialised in web security during development. Working with developers already accustomed to the use of normal security protocols will improve the security of your project and your service. This is because having a good knowledge of the rules of cybersecurity will mean you implement them during the early stages of development. Present the project in production, already prepared for the implementation of the right security protocols.</p><p>Prioritize security and security features of the device before sending it to production. This will protect not only the end user but also the company itself from possible future security problems. Another important factor in ensuring the security of an IoT system is to ensure constant updates, for the entire lifecycle of the device. Moreover, it’s very important to ensure that the device updates automatically without requiring user intervention. Eliminating the need for manual intervention for the update allows you to not only improve the user experience but also to prevent the user from postponing the update.</p><p>Security obviously goes through cryptographic algorithms. Use data encryption algorithms during communications to and from the device and to store passwords; it’s key to ensuring that user data is kept and stored in a stable and secure manner. For example, you can create web apps for your IoT device which always use the HTTPS protocol, so as to protect its users with an SSL or TLS certificate.</p><p>Although it is possible to implement a vast array of protocols, one of the main security risks remains the user themselves. That’s why you have to oblige the user to set a strong password, to change it at least once a year and to activate the two-factor verification. The development team will also have to implement and use smart methods to recover login credentials.</p><p>Once you get to the stable version of the software, it is good to perform, as well as during the whole life cycle of the device, various penetration tests to ensure that the implementing security protocols work well.</p><p>In addition, a company that makes IoT devices should never use a “default” password in the factory settings. This is because users often do not change the password if not forced. Leaving a standard password on the device is like leaving the door open at home day and night. Another fatal mistake that a development team can make is to implement secret and undocumented remote access methods such as backdoors. This hidden remote access does nothing but make the attack platform wider and, if discovered by a hacker, can lead to intrusions against which you can not defend.</p><p>At the end of his talk, Guy Rombaut showed us how easy it is to get the password of a device that has not been designed with the right security protocols. He looked for a simple target like a public webcam and, after finding the public IP, he used logic and an open source tool, nmap, to understand the username and password to access the webcam control panel.</p><p>Nmap can be used to discover hosts and services on a computer network, thus building a “map” of the network. To accomplish its goal, Nmap sends specially crafted packets to the target host(s) and then analyses the responses. For example, Rombaut scans the network in this way:</p><blockquote>nmap -T4 -A -v 192.168.0.0/24 -p 21, 75–100,443, 554, 22222, 3322,3333, 5000, 5550, 8080, 8081, 100002</blockquote><p>With this simple command, he scanned the network, looking for a web server that queried then returned the password and the username set by default (admin/1234). But it is also possible to exploit the simpler bugs of JavaScript code on the access page of these devices. For example, by connecting to a web page using the browser’s developer tools, you can invoke the GetCurPwd () function and return the corresponding value “admin123345”.</p><p>Now imagine that this system is not a simple public webcam but for your car or the door of your home. Can you comprehend how great the risks of an IoT ecosystem without security standards are?</p><p>So a truly safe IoT system should always have three characteristics:confidentiality, integrity and availability, also known as the CIA triad, is a model designed to guide policies for information security within an organisation. The model is also sometimes referred to as the AIC triad (availability, integrity and confidentiality) to avoid confusion with the Central Intelligence Agency. The elements of the triad are considered the three most crucial components of security.</p><p>In this context, confidentiality is a set of rules that limits access to information, integrity is the assurance that the information is trustworthy and accurate, and availability is a guarantee of reliable access to the information by authorised people.</p><p>In addition, our devices should always be able to be controlled by something that can be directly associated with us, such as another device or biometric values (fingerprints or face ID).</p><p>In conclusion, Guy Rombaut hopes that all developers and users will start to become more aware of the risks of the IoT world and put pressure on companies to implement ever better security systems.</p><p><em>Originally posted in </em><a href="https://www.codemotion.com/magazine/security-in-the-iot-generation-the-guy-rombaut-point-of-view-2603"><em>Codemotion Magazine</em></a><em> (Milan), Credit: Claudio Davide Ferrara.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c6f504e02253" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[OneFit — Culture and Values]]></title>
            <link>https://medium.com/onefit-engineering/onefit-culture-and-values-197acc6695a6?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/197acc6695a6</guid>
            <category><![CDATA[values]]></category>
            <category><![CDATA[team]]></category>
            <category><![CDATA[work-life-balance]]></category>
            <category><![CDATA[culture]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Mon, 21 Oct 2019 17:17:45 GMT</pubDate>
            <atom:updated>2019-10-22T10:04:36.118Z</atom:updated>
            <content:encoded><![CDATA[<h3>OneFit — Culture and Values</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/986/1*RJIZJy6h_hpabxlKxRr8KQ@2x.jpeg" /></figure><h4>OneFit Culture</h4><blockquote>THIS IS WHAT WE STAND FOR</blockquote><p>We compare our culture at OneFit as a professional sports team. Every individual player is unique and has their own capabilities. Every person is a crucial part of the team. You can have the best striker in the world, but without a good defense you would still lose the game. Therefore teamwork is of the utmost importance and you have to have each other’s back in order to win. That way you can be proud of your team and celebrate all victories together.</p><p>Our values are hidden in everything we do and everything we are. It is hard to put that into one sentence, mainly because we have over 15 different nationalities in our team and everyone is unique. We believe that every person attributes to our company culture in their own very specific way. And every person is an asset that adds value to the product that we deliver and what we’re proud of.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*0qhDcuuVdJgCI2bekhnkzQ.jpeg" /></figure><h3>OneFit Values</h3><blockquote>THIS IS WHAT WE BELIEVE</blockquote><p><strong>People oriented</strong> — Our people make our company. We believe in equality, kindness and respect towards all of our colleagues. Diversity is the core of our product and company culture</p><p><strong>Trust &amp; Freedom</strong> — When you are hired we will have trust and faith in you. We believe in ‘’just do it!’’ Act in the best interest of OneFit. We believe in your own responsibilities and freedom. Your manager will give you feedback along the way and you can always adjust or tweak anything later on. So, good idea? Just do it!</p><p><strong>Goal oriented &amp; Proactiveness</strong> — We are dedicated to deliver the best product/service possible. This reflects all throughout the company. Together we work hard to reach our goals. It takes hard work and a lot of love. Therefore we foster a pro-active culture so that we can continuously grow and become a better version of ourselves.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=197acc6695a6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/onefit-engineering/onefit-culture-and-values-197acc6695a6">OneFit — Culture and Values</a> was originally published in <a href="https://medium.com/onefit-engineering">OneFit Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[One.Fit Engineering]]></title>
            <link>https://medium.com/onefit-engineering/one-fit-engineering-e7074efbc675?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/e7074efbc675</guid>
            <category><![CDATA[software]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[fitness]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[research]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Fri, 19 Apr 2019 11:44:07 GMT</pubDate>
            <atom:updated>2019-10-21T16:55:33.850Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/986/1*RJIZJy6h_hpabxlKxRr8KQ@2x.jpeg" /></figure><h3>About OneFit Engineering</h3><p>FitTech — Fitness and Technology</p><p><strong>We are on a mission: OneFit Planet</strong></p><p>OneFit’s mission is to make this planet a little fitter. By helping people to integrate sports into their lifestyle and by offering diversity, flexibility and bringing a social aspect in every workout. We do this by offering a flexible monthly membership for a rainbow of workouts. From swimming to fitness, from yoga to padel, from boxing to pilates all available at your fingertips with our app. It’s hard… choosing from over 60 thousand classes every month.</p><p><strong>Engineer at OneFit</strong></p><p>When the world gets more and more digitised, people are becoming at risk of being physically passive and mentally overloaded. We want to turn the tables and use technology to help our members to stay fit. To get there, we need you on our squad.</p><p>We are growing as a tech start-up into the scale-up phase (envision a scale-up as a startup on steroids ;D). To improve our product and services we need great people to bring the workout revolution into the world. We have a lot of exciting development challenges, especially as a back-ender. That’s why we need you.</p><p>Our platform transmits dozen thousands of bookings and check-ins every hour in different countries, currencies and timezones. We are working on multiple integrations with other booking systems for our partners API. We have real time listings of workout availability and we make it possible for our users to see who else is doing that workout. We are working on efficient algorithms to engage our members and help them to reach their fitness goals. Our code quality standards are high and therefore on it self already pretty challenging. There is much more we want to build and so much more we can improve.</p><p><strong>Research and Development Team</strong></p><p>We are working on a variety of projects like waitlist, churn prediction model, venue api, business intelligence, core stability and much more.</p><p>We believe in motivation over regulation. Developing the good way can only be done by a good team. We believe motivation is as important as skills for a team to be successful.</p><p>Challenge standing ideas; be curious about new technologies and share your best practices because you want to.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e7074efbc675" width="1" height="1" alt=""><hr><p><a href="https://medium.com/onefit-engineering/one-fit-engineering-e7074efbc675">One.Fit Engineering</a> was originally published in <a href="https://medium.com/onefit-engineering">OneFit Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Agile infrastructure using automated CI/CD, Kubernetes, without (much) hassle ]]></title>
            <link>https://medium.com/@guyromb/agile-infrastructure-using-automated-ci-cd-kubernetes-without-much-hassle-42b939e69a05?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/42b939e69a05</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[agile]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Mon, 25 Feb 2019 21:02:30 GMT</pubDate>
            <atom:updated>2021-09-24T09:20:32.003Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Agile infrastructure using automated CI/CD and Kubernetes, without (much) hassle </strong>🚀</h3><p>In this article I will share my experience with K8s eco-system to speed up the development. I will point out the problems we were facing and introduce our solutions.</p><p><a href="https://one.fit/"><em>OneFit</em></a><em> allows you to play around at hundreds of gyms and studios with one flexible membership and all kind of sports. Our product is becoming very popular around Europe which means we have a lot of partners, customers.. and also.. </em><strong><em>a huge dev backlog</em></strong><em>.</em></p><p><strong>Problem 1: lack of autonomy — hard to setup and manage local environment</strong></p><p>You probably say “use <a href="https://www.docker.com/">Docker</a>”, and it is indeed great. In theory. Basically, we have about 10 web/back-end applications and more services, which means a lot of internal communications. It is too much hassle to configure networking. In addition, running many containers might consume intensive memory and CPU for desktops. Developers are forced to either run each app and micro-service and also understand the complex service dependencies in order to able to develop. Also.. Need to tackle databases, SSL, DNS, caching and so on. <strong>This is just too complicated!</strong></p><p><strong>Problem 2: Business is scalable, but not the legacy technical solution</strong></p><p>More customers, more partners = means more load/API requests. Our legacy solution was based on a managed hosting solution with a fixed server spec. Also, since we had specific peak hours it didn’t make sense to pay for expensive resources that we need only for certain times. Therefore, we needed something more dynamic that can scale up or down accordingly.</p><p><strong>Problem 3: We had only one test environment = bottleneck in delivering</strong></p><p>Developers and QA are fighting on which branch should be on test in order to deliver. This was an annoying bottle neck which simply slowed us down. Also, this environment was not equal to production which caused inconsistencies. We had to move away from a single, almost static - QA environment.</p><p><strong>Problem 4: “It worked on my machine”.</strong></p><p>Trying to simplify local dev environment created a mismatch between running production and development (e.g. SQLite vs MySQL, Memcache vs Redis etc).</p><p><strong>Problem 5: Too many manual processes, things might break</strong></p><p>You want to move fast, and you break things on the way. It’s that simple. However, we wanted to break less to have the best user experience for our customers.</p><h3><strong>Solution: Kubernetes + Jenkins CI/CD + Simple in-house developed tooling.</strong></h3><p><a href="https://kubernetes.io/">Kubernetes</a> is an open-source container orchestration system for automating application deployment, scaling, and management. In a nutshell, it completes what was missing in Docker — manage communications/networking, and helps you run all those different Docker containers like magic (almost**). The biggest advantage is that it can run on any bare metal whether it’s your cloud environment (AWS, GCE, Azure etc.) or your own computer (previously using Minikube, now can be done directly via Docker tools) — just give it memory and CPU and it works.</p><p>This allowed us to create environments easily — for development, QA or even just a feature branch (the dream of every CTO?). This solved the bottleneck, and helped testing a specific feature before merging.</p><p>We automated this process inside Jenkins, together with automated units testing, end to end testing to ensure we break less, and deliver faster. We solved all the SSL and wildcard networking using <a href="https://traefik.io/">Traefik</a> as an Ingress controller so this is done automatically as well (on K8s.. of course). Each generated environment is almost fully identical to our production environment.</p><p>A typical flow in this case would be: a developer creates a branch and pushes changes, this triggers Jenkins to create an environment on the cloud (within 1–2 minutes) and developer can rsync from his own machine to this cloud (we built our own open-source watcher which you can find below**).</p><p>When a feature is tested, it can be merged to develop branch. When merged branch is erased we also remove the corresponding environment using a webhook (as a startup we don’t have unlimited resources). When this goes to master we are ready to create a release. When tag is created Jenkins identifies it and this is ready to be deployed to production (a separated Google Cloud project in our case).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/831/1*dkxEgVJ1YKOfufTthNLPeg.png" /></figure><p>This has been a significant boost to our developer productivity by improving the speed and consistency of testing, resulting in delivering a better-awesome product! <em>🚀</em>✨</p><p>** Kubernetes works great. Still, it has his own limitations. Here is a short list:<br> — Lack of option for reusability of configurations — <em>If you copy-paste you are doing something wrong</em>. We solved it using <a href="https://github.com/kubernetes-sigs/kustomize"><strong>K</strong>ustomize</a> (the alternative, <a href="https://helm.sh/">Helm</a> is awesome but it was overcomplicated to setup and we wanted to move fast). It is done in a nice concept of <em>base</em> and <em>overlays</em> which can be dependent on each other and patch specific configurations.<br> — We wanted to create parameters to setup docker tag, namespaces and others for our CI/CD. We solved it by creating placeholders <em>${PLACEHOLDER}</em> with old fashion <em>bash sed</em> search and replace. You can find our deploy script <a href="https://github.com/guyromb/k8s-tools/blob/master/k8s-deploy">here</a>. It also renames the resources to be compliant with docker/k8s requirements (e.g. slashes are not allowed — which was a problem when creating a namespace using branch name feature/dev-x).<br> — Same for <a href="https://github.com/guyromb/k8s-tools/blob/master/k8s-watcher">Watcher+R-sync</a>. We created our own tool to sync between the local repository and the container on the cloud. <br> — Check out for more <a href="https://github.com/guyromb/k8s-tools">k8s-tools</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*uV69fat_3TV1Cv-TYzZMFQ.jpeg" /><figcaption>Kustomize Concept</figcaption></figure><p>This suggested infrastructure might not be ideal for everyone, and it is far from being perfect, but I hope this can give you some new ideas for a relatively quick transition to an improved way of working… What do you think? send your comments and ideas!</p><p><em>Next in this series — how did we save thousands of euros on monitoring applications (using Grafana, Prometheus, Sentry and more). Stay in touch.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=42b939e69a05" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Monolith to Separation of Concerns: Software Architecture & Teams]]></title>
            <link>https://medium.com/@guyromb/from-monolith-to-separation-of-concerns-software-architecture-teams-f66e7290762d?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/f66e7290762d</guid>
            <category><![CDATA[architecture]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[backend]]></category>
            <category><![CDATA[software]]></category>
            <category><![CDATA[frontend]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Fri, 18 May 2018 15:36:23 GMT</pubDate>
            <atom:updated>2019-10-21T17:04:05.506Z</atom:updated>
            <content:encoded><![CDATA[<p><em>There are two terms thrown around a lot in the software development industry: Front-End and Back-End. But how did it all start? How does it affect teams? and is it actually good?</em></p><p>Over a decade ago websites were quite primitive. The majority of websites were powered by a server generating a static page with some content. There were very basic interactions with the user: mainly buttons and forms. The flow was simple as well. For example, the server generates a form with some inputs. The user fills in some information and submits it. The data is posted to the server, and the server generates another new page with a success or failure, based on the input.</p><p>This architecture type I just described called a monolith. For example, PHP, ASP, JSP and so on containing HTML view. Later on we started seeing the use of templates and patterns like MVC, but still, front-end and back-end were combined.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/460/1*nj94xHY0P_tg3tShQFkdhw.png" /><figcaption>A Monolith</figcaption></figure><p>A monolith software is very easy to write and much less complex in short-term, no need to setup APIs or calls them, and therefore the development speed is faster (and even the performance too). It also makes life easier when it comes to SEO and A/B testing. However, since everything is written in one place, it can become a tightly coupled “Spaghetti Code” that might be hard to maintain and scale in the long term.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/772/1*7__RVneppwah_t8VkHEswA.png" /><figcaption>Wordpress Code — Monolith</figcaption></figure><p>Later on, when Ajax became widely supported in browsers, we could finally request data in the background without redirecting the user to another page. For example, Gmail in 2004 — no need to refresh manually to see new emails. This point in time is crucial because that is the point when we start shifting to front-end and back-end developers (instead of just “web developers”).</p><p>However, in order to make a simple ajax call, it was previously required to write around 100 lines of code. The use of jQuery library solved it and we were able to make a request with 1–3 LoC (using the $.post or $.get functions).</p><p>In addition, there were new types of devices coming up in the market — mobile, tablets, smart TVs etc. To support different display sizes it was necessary to have a responsive design and some device-specific features — which increased the use of libraries and frameworks (e.g. bootstrap solves the responsive layout issues). Such front-end technologies made the development process smoother and also created much wider possibilities.</p><p>Moreover, when the architecture is a monolith, it made it difficult and inefficient to support different devices. This increased the use of APIs — one back-end serving a different type of clients and interfaces or systems (Reusability).</p><p>The advantages of separation seems clear. And there are much more: Loose coupling (easier to test separately), can mock responses (faster development Speed), smaller server-load and very important — easier support for CDN.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/576/1*mi7zJF2lNWAZfFkisjUFsA.png" /><figcaption>Seperation of Concerns: Front-End and Back-End</figcaption></figure><p>However, while separating front-end from back-end proved its efficiency, it made the back-end also more complicated because we need to take care of things like API versioning, implementing communication interfaces like REST, SOAP etc. For the front-end we need to handle errors from server, sending, receiving and processing data.</p><p>To solve it properly, it requires much more efforts compared to a monolith. In other words, we need more developers to speed up the software deliveries. And bigger teams might result in more conflicts. And this is also the point when your CI/CD pipeline starts failing (and of course it’s a good thing because it means that the integration tests are working).</p><p>Therefore, when the team scales up it is very important to choose the right architecture/patterns. We can reduce conflicts by making teams more autonomous. I would like to show you few patterns as an example how different architectures can fit better with different teams.</p><p><strong>Micro-Pages Architecture — Non-SPA, Micro-services soulmate</strong></p><p>Each team owns a page. Pages communicate in between them using a session (e.g. cookies, local storage etc).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/626/1*hNJ7hQUllGytwq7n98_kaA.png" /><figcaption>MPA Architecture</figcaption></figure><p>Each page is completely independent and has its own repository and dependencies. For example, you can even use multiple frameworks (Angular, React, Vue.js etc) — I don’t say you want to, but you can (might be useful for migrations). It also means you might have duplicates (e.g. header).</p><p>+ Less dependencies<br>+ Simple integration<br>+ Complete autonomy</p><p>— Duplicates of functionalities<br> — Inconsistency of UX<br> — Poor performance<br> — Sharing data is limited via a session</p><p><strong>Portal Architecture — SPA</strong></p><p>Each team owns a widget. Global functionality is shared via the portal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/846/1*NJObSvjoy-ltEdAOzPfinA.png" /><figcaption>Widgets and Portal Architecture</figcaption></figure><p>+ No duplicates<br>+ Consistent UX</p><p>— Portal is a dependency<br> — Portal need to be mocked<br> — Versioning of dependencies</p><p><strong>Summary</strong></p><p>In conclusion, I introduced the monolith first, and moved towards separation of concerns, to more modular architecture that enables large teams to work together on large projects more efficiently. The organizational structure goes hand in hand with Architecture and each pattern has his advantages and disadvantages.</p><p>Modular patterns and micro-services/pages are a trade-off. They might be a good fit for large-scale projects but not necessarily for a startup environment/smaller projects. It might over complicate things and slow down the software deliveries without enough justification. Start with a monolith, separate when it’s time.</p><p><strong><em>Guy Rombaut</em></strong><em> is a founder of </em><a href="http://leagile.co/"><em>LeAgile.co</em></a><em>. We have a proven track record of helping companies realise their digital transformations and speed up their software deliveries. </em>🚀✨</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M7LjypTbPLwMZZWmP7isoA.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f66e7290762d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Security in the IoT generation]]></title>
            <link>https://medium.com/@guyromb/security-in-the-iot-generation-289252d701f8?source=rss-4ae0aed9dc32------2</link>
            <guid isPermaLink="false">https://medium.com/p/289252d701f8</guid>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[internet-of-things]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[backend]]></category>
            <category><![CDATA[security]]></category>
            <dc:creator><![CDATA[Guy Rombaut]]></dc:creator>
            <pubDate>Tue, 14 Nov 2017 13:30:19 GMT</pubDate>
            <atom:updated>2019-10-21T22:23:00.421Z</atom:updated>
            <content:encoded><![CDATA[<h3>Internet of Things — Security and Privacy</h3><p>We are all familiar with the term “Internet of people”. Every minute there are around 4 million posts being shared in Facebook, around 0.5 million tweets, 100 hours of new videos being uploaded. You might even share read this article via one of the social media. This is no doubt that this was a successful revolution.</p><p>Nowadays we are in the revolution of “Internet of Things” (IoT). Basically, “Things” can communicate between them. For example, Lighting system can dim and adjust the lights in your living room for a nice atmosphere, cars are able to be self-driven and so on. The estimates are that by 2020 will be around 100 billion IoT devices being used (account for nearly half or more of connected devices).</p><p>The growth of Internet of Things (IoT) in our daily life creates immense opportunities and benefits for our society. However, IoT security has not kept up with the same rapid pace of innovation and development. This situation creates substantial security flaws and putting our privacy at risk. In the last year, the news published news about security breaches, confidential documents that were stolen, hospitals and power plants being shut down due to cyber attacks and so on. Today everyone can be an hacker even without deep technical knowledge; download some script and just run it (so called script-kiddies). “Things” that are suppose to protect us (home security camera) are opening doors for new types of thieves. This happens because software is not secured good enough.</p><p>As engineers, we should emphasise security from day one. Or in other words, secure by design. In order to make IoT more secured we have to re-think or focus about few elements.</p><p>When we build an IoT device we need to consider 3 main pillars.</p><ol><li><strong>Confidentiality </strong>— We have to allow access only to data or functionality for which the user is permitted. That means use of better authentication. For example, authenticate using two factors — something that the user “know” (password) and something that the user “have” (send SMS to a phone). At the moment many devices use default username and password which are rarely changed. Second one is</li><li><strong>Integrity </strong>— We need to ensure data is not tampered or altered by unauthorized users.</li><li><strong>Availability </strong>— We need to ensure systems and data are available to authorized users when they need it.</li></ol><p>This is especially important because we are also moving to edge of distributed systems. There is too much data being collected and the cloud solution cannot process it efficient and fast enough. Consider a self-driving car which collects images for navigation, to send it to the cloud, process it, and send it back it takes too long. So each device will be able to manage micro processes by itself. Therefore the device itself will have to be protected and not only the cloud.</p><p>Another important aspect is securing IoT networks — IoT network security is a bit more challenging than traditional network security because there is a wider range of communication protocols and standards. We need to protect those new end-points using traditional methods, but also create new innovative ways. For instance, we can monitor the network and detect suspicious behaviours using artificial intelligence.</p><p>We are living in exciting times. Things that till recently would be possible only in hollywood exists in real life. However, companies and engineers have to take more responsibility and focus on protecting users and their data, together with better regulations. Not because it is easy, but because it is necessary.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wOKm9vGQjCy4lLwbpCngAQ.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=289252d701f8" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>