33rd Degree Conference notes – Day 2

Inspiration over Perspiration: Java App Deployment for the 21st century, Axel Fontaine:

• deploying machine image instead of deploying application artifact on server stack,
• there is problem though – machine images are too big to send them over the net,
• running servers in production should be like backpacking – take only bare minimum that’s needed to run application,
• bare minimum set: application, application server, language, libraries and OS kernel (that’s bootable pack, up to 100MB),
• alternatives: Iaas/Paas, Docker/Packer, OSV by Cisco (not production ready though),
• boxfuse.com:
  • easy to start -> boxfuse up war file,
  • runs on VM using e.g. VMWare,
  • nice support to track running images,
  • easy way to run the image in the cloud – but requires reasonable and stable network connection
• logging in the cloud – logstash, logentries,
• advantages: focus on solution application, promote images – do not do patching, less is more – better minimum, what is tested is what i want to run,
• drawbacks: adds complication orchestrating couple of dedicated services.

Patterns for Organic Architecture – let it grow!, Jarosław Pałka:

• architexture is a process,
• you can’t reason about what you can’t measure,,
• sow and cherish
  • capture data about system:
    • find stable parst of system in SCM
    • Michael Feathers quadrant: uglystables, designflaw, tools and breedinggrounds,
    • find fragile parts of the system,
    • show hidden dependencies of the system,
  • refactoring:
    • not always good :),
    • is it worth it?,
    • technical debt -> give it a MEANING,
    • don’t ask for permissions,
  • visualise – but not everything, focus on limited number of metrics,
• sow and harvest
  • modularization into stable parts of the system,
  • set up rules to split the code,
  • no big design upfront (step by step),
  • simply you don’t know every answer to question,
• composting
  • complexity will arise – it is still growing,
  • know how users use application,
  • know features that are in use,
  • tools to track the code (logs, instrumentation, aspects, bug tracker, byteman, support people),
  • don’t use expensive tools – you are the developer, use your creativity,
  • do not comment out the code – just throw it away (scm will remember),
• gains: shorter feedback!
• do not sacrifice resilience of the system for developers’ short-term productivity.

Hazelcast 3 – In-Memory DataGrid, Christoph Engelbert:

• Collection API, Concurrency API and map reduce,
• data partitioning and distribution – automatic discovery of nodes through multicast,
• executor service to run task (one can specify node to run a task),
• SQL like queries,
• transaction support,
• link to project.

How to get started with Scala with examples, Jacek Laskowski:

I haven’t seen the presentation from the beginning. However I had opportunity to see intro to Scalania project. It seems to be a really good thing for people wanting to learn Scala. I’m going to check out the project and work out the exercises. Here is the link to Jacek’s project.