What I have learned from mircoXchg 2015

We attended the microxchg 2015 from 12.2.-13.2. with a crew of 7 people and did not regret it. 🙂
Location was the Kalkscheune in Berlin, a lofty-style industrial building. Nice atmosphere, great catering.

What did I learn?

• Microservices are written in English without hyphen. Thanks Sam! 
• It is interesting to focus on one specific/narrow topic and listen to talks with different perspectives/aspects 
  • But a total of 26 sessions in two tracks did result in some repetitions. 
  • Friday evening nobody wanted to hear ‚docker', ‚conway', ‚microservice‘ anymore.  (Unfortunately our talk was the last one…)
• There is no concise/sharp definition of what microservices are. (‚I know it is a microservice architecture when I see it‘)
  • But most people agreed on these properties: 
    • „Can be deployed independently“
    • „Loosly coupled service oriented architecture with bounded context“
    • „Self-contained and independent runtime process“
  • Unclear are the following questions:
    • How large is micro?
    • Does the UI belong to microservices? If not, where should the UI be consolidated? 
    • How do you handle service to service communication: sync/async 
• Microservice architectures are not trivial.
  • They result in „Complex, distributed, interconnected systems“ (Uwe Friedrichsen) and show emergent behavior. 
  • Technically there are a bunch of challenges you have to tackle:
    • developing, running, deploying, testing, debugging, tracing, discovering, observing…
  • For each of these challenges promising solutions are popping up, but if you have to cope with all challenges at the same time, this might get overwhelming. 
• Clearly microservices are now deep in the technology adoption cycle and are rapidly gaining speed.
  • In some sense it has become the focal point of the agile, lean, devops, domain driven design, container, cloud, continuous delivery movement - supporting each other in a positive feedback setup.
  • If you get it right, then this leads to adaptive (product) organizations that can react extremely fast to (market) changes.
    • If you do not need this: Then maybe you should not go micro.
  • First figures/analysis on the performance of microservice based organizations are available (Adrian Cockcraft): '10 times faster reaction times at 70% cost.'
• And of course as usual: There are no silver bullets.   

What can I recommend?

• Uwe Friedrichsen gave a great introductory talk 
• Same is true for Sam Newman 
• Fred George presented the challenges you have to face with microservices architectures
  • His asynchronous services example pushed ‚loosely coupled‘ to the extremes. Had a nice chat with him afterwards, what this means for product organizations and product managers: 'Command and control‘ is so dead 🙂  
• On the more technical side I definitely enjoyed Jörg Pfründers talk about testing of microservices
  • Great introduction of what it means for your continuous delivery pipeline if you go micro.
  • The slides are in German but the drawings are quite self-explanatory.

• For me the highlight of the conference was Adrian Cockcraft’s State of the Art in Microservices talk
  • "Cost and size and risk of change reduced“ AND "Rate of change increased"
  • With his insights into the Valley’s industry he showed the big picture and why these are so exciting times to live in. 

And finally: 

At leanovate we have built a simple shop application („Microzon“) based on a bunch of microservices (product, cart, billing, ….).

In this lab environment we experiment with different implementation variants, frameworks, infrastructure components to cope with the above mentioned challenges: Microzon on Github

Our talk presents some of our learnings.