In memory databases

I wrote earlier in my post about JFall 2006 about a talk I attended about in memory databases. For many web sites in memory databases are an important part of the software infrastructure. They are essential in making large sets of information accessible.

There are two strategies here, caching and full access. With caching you retrieve information that is needed from a database and store it in a cache. With a cache you cannot count on any information being available. With full access the strategy is to store parts of your database in memory with the option of doing searches and selections based on all database. Condition is that you need some preload mechanism that retrieves the information and makes sure everything is there.

My experience is that in memory databases tend to maximize somewhere around 1 Gb for Java environments. So this also puts an upper limit on the amount of data you can hold and use. Unless you are Amazon and can afford to build distributed in memory databases.

Interesting to see that in memory databases are now available as products. See for a list of vendors CMS Watch, but there are more smaller specialized players in this market.

If you are not ready to invest or cannot find one that suits your taste, it is pretty easy to build your own in memory database. A Collection derived object with some synchronized calls to ensure thread-safe operation gets you underway. Prevayler sounds to me like a good alternative.

Advertisements

One Response to “In memory databases”

  1. buy cialis online Says:

    buy cialis online

Comments are closed.


%d bloggers like this: