Reviews: Memcache
March 19, 2009 at 2:21 pm 1 comment
Is data retrieval from database, killing your performance? Is your database crashing every few weeks, due to load? Does most of your time go into tweaking and tuning the database for performance? Are you spending too much on Db administrator? Do you find Mysql replication an impossible job?
If the answer is yes for any one of the questions above, then you should give a look at Memcache. Memcache is an object caching system. It caches all your db objects to be used by n number of threads or any part of your application or any process across the data centre.
Assuming you have product, which runs on several servers (S1, S2, S3… all of them on the same data centre), you need to optimise it for data retrieval. A Memcache client will have all your key:pair list hashed on it, it can also have all the information about the servers you are using for balancing the load. Now if a request is redirected to one of the servers, S1 and Memcache is installed, then it will look for the object in memcache client, C1. If an existing key pair exists for the object, the obj is returned from the server (any of S1, S2, S3 …) to the S1. If it does not exist, then it will be created and saved in S1 cache.
More documents on the topic.
http://www.danga.com/memcached
http://blog.twitter.com/2008/05/its-not-rocket-science-but-its-our-work.html
http://www.facebook.com/note.php?note_id=23844338919
Like this:
Entry filed under: analysis, Open source, Reviews, software development. Tags: algorithm, Memcache, scaling, tools, video ad network.
Reviews: SVM IDG launches Tech Adnetwork in India.
1 Comment Add your own
Leave a Reply Cancel reply
Trackback this post | Subscribe to the comments via RSS Feed
1.
Bret Clement | March 19, 2009 at 10:19 pm
This is thoughtful review of memcached, thanks for publishing it.