<?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 Rob Davies on Medium]]></title>
        <description><![CDATA[Stories by Rob Davies on Medium]]></description>
        <link>https://medium.com/@rajdavies?source=rss-49b3565698ce------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*_lhx7co8ImLmswiyVjia6g.jpeg</url>
            <title>Stories by Rob Davies on Medium</title>
            <link>https://medium.com/@rajdavies?source=rss-49b3565698ce------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 17:28:42 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rajdavies/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[Distributed Rabbit Holes …]]></title>
            <link>https://medium.com/@rajdavies/distributed-rabbit-holes-abd805d2be5c?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/abd805d2be5c</guid>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Thu, 02 Jul 2020 12:27:23 GMT</pubDate>
            <atom:updated>2020-07-02T12:27:23.524Z</atom:updated>
            <content:encoded><![CDATA[<h3>Distributed Rabbit Holes …</h3><p>One of the great things about being an Architect is that researching a particular topic can lead you down unexpected rabbit holes. At <a href="https://www.cloudbees.com/">CloudBees</a> we are extending the envelope to its limits for GraphQL as we look at having a highly scalable, federated, robust implementation for our <a href="https://www.cloudbees.com/software-delivery-management">Software Delivery Management</a> products. Our implementation also needs to support dynamic schemas and be extremely secure. Throw in how you do rate limiting and circuit breaking, and you have a big bag of big challenges to noodle on.</p><p>One particular rabbit hole I recently fell into was investigating highly distributed consistency and in particular <a href="http://www.cs.umd.edu/~abadi/papers/calvin-sigmod12.pdf">Calvin</a>, which is authored by Dr Daniel Abadi. I first came across Dr Abadi a decade ago when he wrote a <a href="http://dbmsmusings.blogspot.com/2010/04/problems-with-cap-and-yahoos-little.html">blog</a> on why CAP theorem is insufficient. Dr Abadi ended up authoring <a href="https://en.wikipedia.org/wiki/PACELC_theorem#cite_note-ctmddsd-3">PACEL</a>C theorem, which is built on CAP, and states that even when a distributed system is running normally in the absence of partitions you still need to choose between latency and consistency.</p><p>Calvin relies on a deterministic ordering guarantee to significantly reduce contention costs usually associated with replicated systems, and is the inspiration for the global GraphQL data store FaunaDB. The only slight limitation with Calvin is that in order to implement deterministic transaction processing, Calvin requires the preprocessor to analyze transactions and potentially “pre-execute” any non-deterministic code to ensure that replicas do not diverge.</p><p>Although we won’t need to use Calvin just yet, it was fun working out how to implement it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=abd805d2be5c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Making Cloud Native simple: Jenkins X]]></title>
            <link>https://medium.com/@rajdavies/making-cloud-native-simple-jenkins-x-3272d4362bef?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/3272d4362bef</guid>
            <category><![CDATA[docker]]></category>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Tue, 20 Mar 2018 04:16:50 GMT</pubDate>
            <atom:updated>2018-03-20T04:16:50.992Z</atom:updated>
            <content:encoded><![CDATA[<p>You have probably heard that all companies are now software companies and that to compete you need to be embracing cloud native applications to ensure your company has the ability to adapt quickly — or you will soon be out of business. This is true by the way, but its also a little daunting for all of us. You have to be an expert in your programming languages of choice, be well versed in how your business works, and now need to understand what exactly <a href="https://www.cncf.io/about/faq/">cloud native</a> means, what are containers and what are the latest trends in cloud technologies.</p><p>So lets make a quick list of things you should be doing:</p><p>1. Decompose your monolithic applications — or not. To generalise, you probably really should give this a lot of careful thought: its domain and application specific — i.e. only you can decide if it make sense (but it probably does).</p><p>2. Containers — or <a href="https://medium.com/devopslinks/an-overall-view-on-docker-ecosystem-containers-moby-swarm-linuxkit-containerd-kubernetes-5e4972a6a1e8">Docker or Moby</a> — which really you need to think about <a href="https://containerd.io/">containerd</a>, or <a href="https://github.com/rkt/rkt">rkt</a> or something that complies with the <a href="https://www.opencontainers.org/about">OCI</a>. Which is why people talk about containers — its all got too confusing.</p><p>3. Orchestrating your containers — i.e. service discovery, auto scaling, fault tolerant platform to run your containerised applications. Thankfully, this recently got a whole lot easier — with <a href="https://kubernetes.io/">Kubernetes</a> emerging as the <a href="http://www.actoncloud.com/blog/kubernetes-k8s/">one</a> that will be supported by all the major public clouds, and existing cloud platforms.</p><p>4. Configure, release version, rollback and inspect your application deployments in a standard, or de facto standard way. Luckily <a href="https://github.com/kubernetes/helm">Helm</a> is excellent for this.<br>5. CI/CD — in order the extract the most value from cloud native, you need a continuous delivery system that will enable a predictable, repeatable release process, and enable continuous improvement via streamlining your development processes.<br>6. Monitoring — ideally you want to monitor the performance of your deployed applications and feed this back into your CD system.</p><p>This is an intimidating list to look at, but its probably already out of date because technology is moving at such a pace its really difficult to keep up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/200/0*5rDyG7PAi9dcdSLD." /></figure><p>What if you could abstract yourself away from all the technology bleeding edge and save yourself the paper cuts — and just concentrate on adding business value? Well this is the aim of <a href="http://jenkins-x.io/">Jenkins X </a>— a new project that is part of the Jenkins ecosystem, as explained by James Strachan <a href="https://jenkins.io/blog/2018/03/19/introducing-jenkins-x/">here</a>.</p><p>We started Jenkins X at the begining of the year, taking the experience gained from the <a href="https://fabric8.io/">fabric8</a> project to develop an open source system that was targeted at these aims. By concentrating on Kubernetes and utilising its wider ecosystem at the <a href="https://www.cncf.io/">Cloud Native Computing Foundation</a>, we have been able to develop a robust, targeted project that focuses on the needs of developers of cloud native apps.</p><p>In summary Jenkins X provides the following:</p><p>1. Abstracts away the gnarly bits of cloud native (you probably don’t want be concerned with using <a href="https://github.com/GoogleCloudPlatform/skaffold">skaffold</a> for example). Its all still there, you can peel back the curtain as much as you like, but its nicer if you don’t have to.<br>2. Automated CI/CD Pipelines — using Jenkins, configured to work well for cloud native<br>3. Environments — promotion using git based work flows, and preview environments for pull requests.</p><p>This is all setup for you — you can aim Jenkins X at an existing project (or create a new one from scratch) — and select your cloud provider of choice and let us do the rest.</p><p>A small(ish) caveat: we target Kubernetes, so check <a href="https://kubernetes.io/docs/setup/pick-right-solution/">here</a> first.</p><p>Most public cloud providers allow free trials, or provide credits — which is more than enough to kick the tyres with Jenkins X — or you could try a local-machine solutions, such as <a href="https://kubernetes.io/docs/getting-started-guides/minikube/">Minikube</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3272d4362bef" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The fabric8 project keeps on growing]]></title>
            <link>https://medium.com/@rajdavies/the-fabric8-project-keeps-on-growing-352e3b23f4f?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/352e3b23f4f</guid>
            <category><![CDATA[continuous-delivery]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Tue, 02 May 2017 15:14:38 GMT</pubDate>
            <atom:updated>2017-05-02T15:14:38.815Z</atom:updated>
            <content:encoded><![CDATA[<p>The <a href="https://fabric8.io/">fabric8</a> project has been at the forefront on innovation to enable software professionals to develop and deploy faster to cloud native environments. With t<a href="https://www.redhat.com/en/about/press-releases/red-hat-unveils-end-end-cloud-native-development-environment-red-hat-openshiftio">he announcement of openshift.io</a> at Red Hat Summit, 2017, the eco system of fabric8 has expanded to incorporate all (except the IDE — which is based on <a href="http://www.eclipse.org/che/">Che</a>) of the technologies that make up the <a href="http://openshift.io/">openshift.io</a> developer platform:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/585/0*Xy-t_J98Co0N7JAP.png" /></figure><p>The fabric8 project will continue to innovate and increase its scope to ensure it becomes the best environment for developers to continue to accelerate from idea to production. The fabric8 platform consists of over a hundred repositories under the <a href="https://github.com/fabric8io">GitHub fabric8io organization</a> — all built on top of the fabric8 platform itself, keeping all releases in sync, and allowing us to continually improve our delivery.</p><p><em>This post was originally posted on</em><a href="http://www.rajdavies.today/"><em> rajdavies.today</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=352e3b23f4f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Moving anything towards a container based, Microservices architecture is pretty hard.]]></title>
            <link>https://medium.com/@rajdavies/moving-anything-towards-a-container-based-microservices-architecture-is-pretty-hard-5d7a19e16be8?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/5d7a19e16be8</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[docker]]></category>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Sat, 24 Sep 2016 07:55:40 GMT</pubDate>
            <atom:updated>2016-09-24T07:55:40.440Z</atom:updated>
            <content:encoded><![CDATA[<p>Moving anything towards a container based, Microservices architecture is pretty hard. Doing it while the technology has been evolving so quickly is a even harder, but creating a user experience that means Microservices is really easy to build is the hardest thing we’ve ever done — and we would love to tell you about it!</p><p>If you are in Atlanta or Minneapolis the first week of October, Red Hat is hosting <a href="http://microservicesjourney.com/">A Microservices journey with Apache Camel</a>, where James Strachan, Claus Ibsen, Christian Posta, James Rawling and myself will tell you about the expedition through containers, kubernetes/<a href="https://www.openshift.com/">OpenShift</a>, continuous delivery, API management and how we’ve made it all super easy to use with <a href="https://fabric8.io/"><strong><em>fabric8</em></strong></a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5d7a19e16be8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Fabric8 MaaS: Scalable Messaging as a Service]]></title>
            <link>https://medium.com/fabric8-io/fabric8-mq-not-just-a-message-broker-18ca5af1e0f?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/18ca5af1e0f</guid>
            <category><![CDATA[activemq]]></category>
            <category><![CDATA[fabric8]]></category>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Mon, 22 Jun 2015 11:26:12 GMT</pubDate>
            <atom:updated>2015-07-02T14:39:42.607Z</atom:updated>
            <content:encoded><![CDATA[<p>Enterprise message brokers like Apache ActiveMQ are designed to support many different patterns of messaging, from point-2-point, publish/subscribe and many flavours in between. They are resilient, support high availability, automatic failover, store and forward and different protocols from MQTT, STOMP,OpenWire, AMQP to WebSockets. Over 10 years of development, ActiveMQ has become extremely feature rich and flexible, however this flexibility can have some performance cost.</p><p>However, traditional message brokers are not designed for the type of flexibility required for cloud: you need very high scalability and the ability to scale up and down the number of message brokers based on demand. The number of concurrent clients a message broker can support isn’t an exact science: it depends on what the clients are doing — how many destinations (Queues/Topics) are being used, and how much message load each client will produce. As a rough guide — the upper limit is going to be 10,000 clients at best.</p><p>ActiveMQ supports high availability through a master/slave mechanism, where there is only one active member of a cluster and one or more passive slaves. In such a scenario, you can scale up and down brokers, but only the master will be active, so its not going to help horizontally scale up your message brokers.</p><p>ActiveMQ also supports store and forward (or networks of brokers) — where messages are delivered from the client to the connected broker, an forwarded on to any other broker that has interested consumers for the messages. This is designed for passing messages over a WAN, but it can also be used to form a mesh of interconnected brokers on the cloud. The benefits are that this could provide a mechanism for scaling up and down brokers. There are some downsides though:</p><ul><li>Networks are chatty — routing is defined for messages based on where message consumers live. In order for networks to work, information is passed from broker to broker to so routing can be determined.</li><li>Messages may have to perform more than one hop to be routed from producer to consumer. This introduces extra latency, and extra load across brokers.</li><li>Messages can be orphaned. If a broker goes down, it could have locally stored messages that won’t be delivered until its recovered. You can mitigate against this by having master/slave brokers for each vertex in your interconnected mesh. However, this also introduces extra passive brokers that will sit around pondering their existence as a slave.</li></ul><p>Ideally, to achieve performance and scalability, you need to have destination sharding: the ability to group destinations and their associated consumers and producers on the same broker. This not only reduces the latency and number of network hops, but also allows ActiveMQ to make some internal optimisations when dispatching messages, without impacting integrity of persistence.</p><p>The flexibility that ActiveMQ provides is also a barrier to performance and scalability: The more layers you have between ActiveMQ storing and routing messages, the worse its performance is going to get. If you want ActiveMQ to be automatically scalable, you would almost need to start again — and this is what <strong>Fabric8MaaS</strong> is — a way to use existing message brokers (ActiveMQ) and make them scale horizontally.</p><p><strong>Fabric8MaaS</strong> isn’t even a message broker, its a smart messaging proxy, that utilizes Fabric8/Kubernetes to create Messaging as a Service out of automatically sharded ActiveMQ brokers. It sits on top of a <a href="http://vertx.io/">Vert.x</a> core to provide highly scalable asynchronous connection handling:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*40y33EILe7jUwM9dpqxUpQ.png" /></figure><p>Its best explained by following what happens to messages as they flow through Fabric8MaaS Proxy:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/792/1*BexvR2R5TyWwJ0mfgpqMEg.png" /><figcaption>Fabric8MQ Message Flow</figcaption></figure><ol><li>Protocol Conversion — by moving MQTT,STOMP,AMQP protocol conversion outside of the ActiveMQ broker reduces the amount of work the broker needs to do: As ActiveMQ isn’t well designed for modern, asynchronous tasks, the less work the broker does, the better.</li><li>Camel Routing is built right into the router — so we can easily convert on the fly between, say Topics (Publish/Subscribe) and Queues (point-2-point).</li><li>API Management — utlizing <a href="http://www.apiman.io/latest/">API Man</a>, so you can apply security and rate limiting policies to destinations</li><li>Multiplexer — reducing the amount of physical connections an ActiveMQ broker has, increases the throughput of the overall system.</li><li>Destination Sharding — this is where a lot of the magic pixie dust is used, to shard connections across many different backend ActiveMQ brokers. Its also where messages and clients are moved between brokers, as brokers are spun up and down, depending on load</li><li>Broker Control — this monitors the brokers under Fabric8MaaS control and monitors load — and it decides if more ActiveMQ brokers are needed, or less — depending on load. An embedded rules engine is used to make the decisions.</li></ol><p><a href="http://fabric8.io/guide/fabric8MQ.html">Fabric8M</a>aaS is <em>experimental</em> and part of the amazing <a href="http://fabric8.io/">Fabric8</a> project.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18ca5af1e0f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/fabric8-io/fabric8-mq-not-just-a-message-broker-18ca5af1e0f">Fabric8 MaaS: Scalable Messaging as a Service</a> was originally published in <a href="https://medium.com/fabric8-io">fabric8 io</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Microservices, API Management and Fabric8]]></title>
            <link>https://medium.com/fabric8-io/microservices-api-management-and-fabric8-bf17b453b358?source=rss-49b3565698ce------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf17b453b358</guid>
            <dc:creator><![CDATA[Rob Davies]]></dc:creator>
            <pubDate>Wed, 06 May 2015 11:34:50 GMT</pubDate>
            <atom:updated>2015-05-06T11:34:50.391Z</atom:updated>
            <content:encoded><![CDATA[<p>“Microservices” highlighted by Martin Fowler and James Lewis <a href="http://martinfowler.com/articles/microservices.html">last year</a>, have set in motion a huge amount of excitement and hype around a new software architecture — grown out of the need to provide highly scalable deployments for cloud enabled environments. There has also been a lot of talk about the death of the good old Enterprise Service Bus (ESB)- in fact, if you search for ESB is dead or ESB is DOA, you’ll see a lot of articles over the last 10 years saying that either the ESB killed SOA or its now an outdated concept, and you should be thinking about the API economy.</p><p>Microservices are really about doing SOA the unix way, having single purpose services that can be independently scaled if needed. This also follows on from following the design principles of <a href="http://12factor.net/">twelve factor apps</a>.</p><p>However, not all architectures are going to be suitable for the new way of doing things, and if you are responsible for an existing architecture, you may not be in able to suddenly drop everything and run after the latest trend.</p><p>One of the biggest problems to date with microservices is having the right infrastructure in place to discover, manage, monitor and deploy your services (be they nano,micro or monolith). If you have the right software in place, it suddenly becomes really simple to apply best practices.</p><p>This is why we built <a href="http://fabric8.io/">fabric8</a>. If you’d like to find out more about microservices, API management, integration and the best way to glue them altogether, and talk to some of the actual developers behind fabric8, <a href="http://vertx.io/">Vert.x</a>, <a href="http://camel.apache.org/">Apache Camel</a>, and much more — there’s a Microservices Architecture day that’s going to be popping up in London on June 11th: more details to follow.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf17b453b358" width="1" height="1" alt=""><hr><p><a href="https://medium.com/fabric8-io/microservices-api-management-and-fabric8-bf17b453b358">Microservices, API Management and Fabric8</a> was originally published in <a href="https://medium.com/fabric8-io">fabric8 io</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>