There’s been a ton going on in April, and I wanted to get some of the highlights down in a post (in no particular order):
After about 3 years of eagerly anticipating the launch of a stable Redis cluster, it has finally landed in Redis 3.0!
We run several clusters of Redis and push it pretty hard. Until now, we’ve been manually sharding our data across standalone Redis instances, dreaming of the day when Redis is capable of operating in a self-contained distributed system. Finally that day has arrived!
We didn’t waste any time in migrating our sharded setup to a proper cluster. Way before Redis cluster even hit its stable release, Simon started working on adding support for Redis cluster to his redis-clustr node module. As of this month, redis-clustr is the most stable and fully-featured node.js module for Redis cluster that we know of, and we’re currently using it in production with our shiny new redis cluster setup.
I mentioned in a post that we were attending the AWS Summit in London this month. AWS invited us to speak as a guest customer in the Keynote, as well as in the showcase startup session later in the day. Take a look at the write-up of our day that I published on our non-techy blog – it has funny photos of James and I waving our arms around on stage.
New AWS announcements
We’re pretty intrigued by some of AWS’ announcements this month:
* Amazon Machine Learning service could have huge potential. When we get a spare moment, we’re going to put it through its paces and find out what kind of cunning (or dumb) models we can conjure and train!
* SSD EBS can now be sized up to 16TB, up from the previous limit of 1TB. That will make our Cassandra nodes happy.
As if his foray into distributed systems with Redis cluster wasn’t enough, Redis creator Salvatore (Antirez) has launched himself into a new side project: Disque, the code for which was published this week. In short, Disque is a distributed message broker.
We’re incredibly excited by this project. Scaling message queues is pretty hard – there are options out there, such as Amazon Kinesis and Apache Kafka, but we’ve not seen a migration path we’d be happy with yet. Disque looks very promising. Sharing Redis’ pedigree of fast in-memory storage and ability to scale horizontally in an available and partition tolerant distributed system, it just might form the backbone of our internal message stream in the near future.
We’ve migrated most of our main Node.js services to io.js. Their regular release cycle and up-to-date V8 deps mean better performance, stability, and access to ES6 language features that are landing in Chrome. Not long now before we see io.js v2.0.0 which will introduce support for ES6 classes!
— Chris Dickinson (@isntitvacant) April 27, 2015