A relational database in memory?

I have a simple question about Radish. If the key to its performance is that it is in memory, cannot the serum be executed on regular SQL db?

+5
source share
6 answers

Any DBMS can be run "in memory". Consider using ramdisk . However, most DBMSs (with SQL) are not designed to work fully in memory and put a lot of effort into minimizing disk I / O and swap: the DBMS works very hard so that the "relevant data" is hot (in memory and in cache) - - IO slow, slow slow.

This is because database data is often [and historically] significantly larger than main memory. This and main memory are unstable :-) [ACID DBMSs do a lot of work with future writes) to non-volatile storage - and other methods that ensure that data is never corrupted, even in the event of an unexpected shutdown. ]

Some databases, such as SQLite, use the same format for disk and storage devices, although they explicitly support in-memory storage. Support for other [in-memory] back-end and memory usage settings is vendor-specific.

Happy coding.

+8
source

You may be interested in VoltDB

+5
source

, , SQL DB. Redis , GET, SET ( ..) - .

SQL- , ( ). , , JOIN- Redis,

+2

TimesTen ( Oracle).

11g SQL , , Oracle.

+2

SQL. .

, , . , ACID- ; , . ( , , dbms SQL , . , .)

+1

: :

- [...] - , Moores .

     

, :

     

.

+1
source

All Articles