What is The Best Database for a Souped Up Hashtable?

By Angsuman Chakraborty, Gaea News Network
Friday, July 6, 2007

All I really want is to store a very large Hashtable with more reads than writes. It should be able to deal with more than 8GB of data. The key will be a String of limited length. I want it to be extremely fast on read-access, stable with sudden shutdown and other issues beyond our control and with very low CPU usage. The leading contenders in my mind are:
1. HSQLDB
2. H2 Database, from the creator of Hypersonic SQL which is the predecessor of HSQLDB
3. Berkeley DB (Java version)
4. MySQL (least preference)
5. Derby Database
6. One$DB
7. Home-brewed implementation of a souped up Hashtable
8. Prevayler

Note: The databases are listed in no particular order.

Preferably it should leverage the memory for faster access using an LRU list for example. However it shouldn’t depend on being fully memory driven. To summarize high read performance with acceptable write performance and low CPU usage are my primary concerns.

Looking forward to your recommendations.

Discussion

Indika
April 27, 2009: 11:12 pm

I have the same requirements on the database. Care sharing which database you finally chose? Indika

July 10, 2007: 7:01 pm

Fast reads, low volatility? LDAP servers are built for that kind of thing.

Cheers.

July 10, 2007: 3:47 am

Have a look at db4o . It’s available for Java and .Net

July 7, 2007: 11:32 pm

That is a very interesting product. Thanks for the pointer.


Taylor
July 7, 2007: 10:26 pm

Have you considered Terracotta ? Very large hashmaps are very easy.

YOUR VIEW POINT
NAME : (REQUIRED)
MAIL : (REQUIRED)
will not be displayed
WEBSITE : (OPTIONAL)
YOUR
COMMENT :