July 4, 2008 at 12:37 am
-
alt microblogging platform with a few key wins over Twitter & Jaiku: stability (so far!), open, decentralized, and Affero-licensed OSS. I’m “jm” on it, but not writing there — yet. but looking forward to an API so I can add it to twit.ie
-
some third-party app developers get access to it, some don’t. one dev says: ‘It’s frustrating to just get locked out after spending so much time making stuff for Twitter users’
-
910-node cluster sorting 1TB of data in 209 seconds, using Hadoop and HDFS. I wish we had a Hadoop cluster to do SpamAssassin mass-checks on ;)
-
‘a fast, distributed, in-memory workqueue service’, written in C with libevent, lots of client libs for different languages. Nice lifecycle model. The queues are not persistent yet, though, unfortunately
Permalink
RSS feed for comments on this post
View blog reactions using Technorati
Steve Loughran said,
July 4, 2008 @ 8:27 am
Justin, if you have Petabytes of spam data -and want to work with Mahout or Hadoop core, then get on the list and start talking about what you have and what you need. Someone should be able to sort out some time on one of their clusters, even if isn’t one of the big Yahoo! ones.
Justin said,
July 4, 2008 @ 12:19 pm
Steve — thanks for the tip!
We only have GBs, not petabytes, but we do have a need to re-run our mass-checks (mass scans of GBs of mail using SpamAssassin) over the entire collection on a very frequent — currently daily — basis. I’d say that might be a problem, but I should really get off my ass and get over there to find out ;)