What time is it? It's Devoxx time! [Day 2]
Second day of Devoxx Poland 2016 conference.
Understanding Microservice performance
Rob Harrop - Skipjaq
I've lost my notes and can't remember anything from this lecture. Well, tough morning.
Evolutionary architecture
Neal Ford - Thoughtwork
Second appearance of Neal and after the first one I couldn't miss this one. He expanded ideas he mentioned during his keynote.
When you design your architecture you have to understand the need of change. Lecturer referred to Docker's unbelievable influence on systems architecture. This isn't just an individual case. World of software development is changing constantly, dynamically and incalculably. According to Neal this is one of the main reasons why long term planning is impossible. He proposed an evolutionary approach which primary value is a support for incremental changes.
Of course microservices are still on the top and we also had something about them here. I've never thought about exchanging the whole persistent layer of any software nonetheless layered architecture is still crazy popular. Neal noticed that even though this approach can have multiple dimensions of possible change when we focus on business value smeared all across the system the reality is we are closed for any change. This problem does not affect microservice based systems as they are quite easily replaceable. This trait makes solid base for the evolutionary architecture. Another proposed pattern that could facilitate iterative changes are feature toggles.
Later lecturer mentioned that we should prefer BASE (Basic Availability, Soft-state, Eventual Consistency) over ACID (because world isn't consistent and we are constantly waiting for everything to set). There were also some repeated stuff from keynote like central governance and Conway's Law. Unfortunately organizers assumed that every talk should lat 50 minutes so Neal's one got shortened in comparison to this version.
Reactive Microservices with DDD and Actors
Vaughn Vernon - for{comprehension}
And now's the time for Vaughn Vernor, first man who understood Eric Evan's Domain Driven Design and wrote a book about it. During the conference everyone likes to mention big ball of mud, apparently it looks nice beside microservices (and you get additional points if you mention them).
Vaughn also mentioned ACID and that before the computers everything was eventually consistent. A few words about DDD in a context (wink wink) of microservices and that they should be the size of bounded context. Vernon tried to explain types of microservices (rapids, rivers, ponds), but I drifted away in the moment when he started showing code, taking about Scala, Akka and Play Framework. During this lecture I felt like at the university with professor that is completely devoted to his believes and preaches his wisdom. I like actors and microservices but this preach hasn't reached me.
Data consistency: Analyse, understand and decide
Louis Jacomet - Terracotta/Software AG
I was sure I'm going to another talk about microservices and eventual consistency but I couldn't have been more wrong! Well, there's a possibility that I've never read the description.
Louis talked a bit about ACID and cited some Wikipedia definitions. He notet that applications rarely take advantage of multiple connection pools with distinct isolation levels. I had no idea that cache, for example in Hibernate, could utilize isolation level itself (you use this crap every day and still know nothing). Lecturer brought up essential remark that transactions won't give us certainty if we employ ANSI defined isolation levels as anomalies in relational systems are discovered every day.
Functional data structures with Java 8
Oleg Šelajev - ZeroTurnaround
I wanted to check this panel as I recognized Oleg from vJUG sessions and even the topic grabbed my attention. I really really want to like this talk but I simply can't. Despite nice and substantive slides, lecturers wisdom and best will, something was wrong. He seem a little bit nervous and not really well-received. Those issues gave the feeling of tediousness and dullness.
Oleg showed differences between functional data structures and those commonly known for example from university (envy those who were taught functional data structures in school). Speaker described few optimizations which enable effective operations on immutable structures and also some of their implementations. I got really interested in HAMT (Hash Array Mapped Trie).
How to Create a New JVM Language in Under an Hour
Oleg Šelajev - ZeroTurnaround
Again Oleg's talk and again a letdown, even jokes were the same. To be honest I just wanted to see room number 5 which is located underground, on a level -2 (it looks like some kind of a bunker).
The presentation was more or less summary of Jakub Dziworski's blog about creation of Enkel programming language.
The Pendulum
Bruce Tate - icanmakeitbetter.com
From client forms processed by systems written in COBOL to web applications. Bruce Tate tells a tale of history that repeats, in software development. Pleasant talk, felt like a casual column in your favorite magazine. Bruce for sure likes Elixir language (yay!).
And thats how I finished day two. There were some other short talks and product showcases but I was so tired that I couldn't stand another hour there.