<antirez>

News posted by antirez

Github reviews as a way to improve code quality?

▼
antirez 5724 days ago.
Today I tweeted quite a bit about Ruby software quality. The trigger was simply a Ruby/Sinatra application I run for personal use that crashed from time to time due to a GC bug in the Ruby interpreter I was using (the default of a recent Ubuntu install, sadly).

This is an interpreter bug, and I hope it is a rare one, but this episode made me thinking about how many bad stories I accumulated since my preferred high level programming language is Ruby: bad documented gems, not working gems, dependencies hell, hard to install things, bad performances, and so forth.

Redis Manifesto

▼
antirez 5740 days ago.
Many times I don't know what to exactly reply to feature requests, or questions about why things in Redis are done in a specific way. Most of the time the questions make a lot of sense, there is not just a way to make things in programming, a lot is about your taste, feeling, and ideas about how software should be written. So I tried to condense my feelings about Redis and software development in general in this short manifest, that I'll include in the Redis distribution. There are an infinite number of ways of doing things, this is just the one I and a good part of the Redis community like.

Redis Presharding

▼
antirez 5744 days ago.
Redis cluster is currently under development, and we hope it will be able to solve the problem of partitioning data among different Redis instances in a transparent, fast, and fault tolerant way. The cluster project is currently not ready: even if we have a pretty clear design, and many networking level code for gossip and failure detection, it will take some more month to be released, and more time to be released into a stable release.

While our work continue, people shard regardless of Redis cluster, using their own algorithms and systems. This post describes a strategy for partitioning your data among N instances that is very simple but that works great. We'll call this strategy "Redis Presharding".

there is something between append only and reusing blocks

▼
antirez 5762 days ago.
One of the most interesting properties of an append only btree is that it's impossible to corrupt. There are other interesting properties as well like concurrent accesses are trivial, since whatever is the root and all the nodes your reader is going to access, they are all valid, just they may be an old representation of the btree.

But there is a price to pay that I'm not happy with, that is, the append only btree requires a compaction process, otherwise it will get bigger and bigger. If your use case is mostly-reads this may be ok, but if you plan to use a lot of writes there is some math to do.

Pull requests are not conversations

▼
antirez 5764 days ago.
I don't remember anything that changed my work flow in a so big way as git did in the latest two years. I will not argue too much about that, as you probably experimented the effects yourself if you are reading this blog post, or perhaps you are among the guys I should say thanks for the fact I started using git.

But if git is cool, Github makes it cooler for sure. Github is like a market for git, where code gets exchanged, shared, publicly hacked. In short I love git and github. Still I think that github pull requests are posing a problem in the open source development world, and this post is all about this.

About the new INFO command, Diskstore and Cluster ETA

▼
antirez 5776 days ago.
With this blog post I'll try to keep you posted on the progresses of Redis, the recent changes to the INFO output, and the ETA for diskstore and cluster.

New INFO output
During the week end I managed to improve the INFO output that was rapidly becoming a mess ;) INFO is a very interesting command for Redis users, but there are a few issues that need to be addressed:

• Too much output., not separated by section. This has many drawbacks: complex to read for humans. Time consuming to generate as it can take more than a millisecond sometime, for some Redis uses this is not ideal.

Diskstore and a new b-tree implementation

▼
antirez 5782 days ago.
A few weeks ago I presented a new development direction for Redis, called diskstore. The original announcement is in this Redis Group post.

Diskstore is working very well in our initial tests, and we are working to improve it. Diskstore is basically a smart caching layer on top of Redis backed by an on disk plain key value store. The current implementation of the disk store is file system based: every key is represented on disk by a single file. This setup is interesting in many ways as it is pretty hard to corrupt, and very transparent (you can see the individual files, rsync them into another server, and even joining two databases just merging all the files), but hardly the fastest.

Redis 2.2.0 RC1 is out

▼
antirez 5816 days ago.
Finally we are already at the end of another development cycle: the first release candidate of Redis 2.2.0 is out! You can grab it here from github, it is tagged as 2.2.0-rc1, or if you prefer, as a tar.gz file.

As I said some time ago, 2.2 is a monument to how much Redis 2.0 sucked. I think that looking back to the previous version and having the feeling of a very bad implementation is a good thing. It means we did another step improving our code.

What's wrong with 2006 programming?

▼
antirez 5887 days ago.
Redis 2.0 introduced a new feature called Virtual Memory. The idea is that some applications using Redis may not access the whole dataset with the same frequency. In extreme cases only a little percentage of hot spot data is used often, while the rest is mostly idle and touched very rarely. For instance imagine a Redis instance holding User objects: the most active users will hit this subset of records continuously, while a large percentage of users will access the site a few times a month, and another large subset of users completely forgot about this web service at all.
: