Twitter is experimenting with Redis

github.com

34 points·jrnkntl··4 comments

Twitter seems to experiment with a lot of stuff. A few years ago, there was buzz in the Groovy community because Twitter was experimenting with it. I thought that was cool until I talked to a friend in the Python community, who said Twitter was experimenting with it. A few months later, they announced they were using Scala.

It seems (admittedly, from that one example), that Twitter puts significant investment into exploring things before making a decision. Which seems like a great way to do things.

I would not be overly surprised to find in a month that Twitter is developing something similar to Redis. They do seem to roll their own technology quite a bit. Is it NIH or do they really have that many unique problems (e.g. scale) to solve?

On point, there's Retwis by antirez, which models Twitter to some extent using Redis as a backend.

http://retwis.antirez.com/

Twitter is not in any way unique or hard. If they spent more time thinking and less experimenting, they would have figured that they need a small in memory databse, and would have solved their technical problems easily and cheaply more than two years ago.

Seconded. These twitter stories make me bite the table every time.

It boggles the mind that they are still failwhaling constantly, after all this time.

Why won't someone finally fire the tech team, all of them, and hire someone with a clue?

The twitter application is about as trivial as it gets for a messaging application. Much more complex systems, both distributed and centralized, handle much higher loads every day, without dropping out all the time.

Redis is a fine piece of software, mind you. But one would think they'd be a bit beyond the stage of "experimenting" after four years of operation....

Reply on news.ycombinator.com