In-Memory Database Systems

IMDSes are especially useful for embedded development, where every saved process shrinks the footprint and the bottom line.
______________________

Comments

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Caching from other databases

Kanchana's picture

Nowadays, in memory database systems are being used to handle load for high volume web sites. Amazon uses IMDB technology to handle its massive database load from its web servers. IMDB systems are used for caching tables from traditional disk based systems such as Oracle, DB2, MySQL to improve the performance and maintain the freshness of the data (keep cache in sync with the actual database). Examples are Timesten and CSQL. Timesten provides caching for Oracle database and CSQL provides caching for Oracle, MySQL, Postgres, etc.

Re: In-Memory Database Systems

Anonymous's picture

What happens to In-Memory Database Systems when the database operation results in a large result (say a select operation which selects 5 million tuples from the databases)?

Are In-Memory Database Systems suited only for small databases ?

Re: In-Memory Database Systems

Anonymous's picture

64-bit versions of in-memory databases are less constrained than 32-bit versions, obviously. Another, often overlooked, concern is the amount of physical (versus virtual) memory in the system. An IMDS could have a 1 gigabyte database in a system with 2 gigabytes of memory, but if the size of all concurrent processes exceeds physical memory, the O/S will swap something out.

A 5 million tuple result is not, in itself, a problem. If the database must build a temporary table to create the result set, that could be a problem if available memory is limited.

Like their disk-based cousins, IMDS come in different forms. Your question pertains more to SQL databases. Some IMDS (my company's eXtremeDB, for example) are what are called "navigational", meaning the programmer navigates through the database one record at a time using the database programming interface. There are advantages and disadvantages to both types of programming interface. One advantage of the navigational API is that it is never necessary to build temporary tables because ad-hoc queries are non-existent with navigational APIs, by their nature. (In the absence of ad-hoc queries, where all access paths are known in advance, a well designed database with an SQL interface would also not require a temporary table for any query.)

-Steve