views:

319

answers:

6

as far as I understand postgres, mysql or oracle will be wear out ssd disk very fast.

I need one INSERT query per few second and statistic processing of inserted lines per 24 hour.

platform: linux

A: 

Consider using SQLite, it's small, lightweight, and can be tweaked to lower its impact on the filesystem. do note though, that ensuring atomicity comes at some IO and performance cost. You should always try to batch DB updates together to reduce overhead.

Hasturkun
A: 

It's not done yet, but RethinkDB, an YC startup, is developing a storage engine for SSD's.

Roman Plášil
A: 

Did you try Gemstone? An OODB has different access patterns.

Dale seems to be rather impressed

Stephan Eggermont
Hm. Why GemStone? I don't find anything about compatibility with ssd
vitaly.v.ch
A: 

Firebird work's well on SSD disk.

Hugues Van Landeghem
+1  A: 

There is mentioned above RethinkDB specially designed for the SSD. In the same time there are access patterns where existing DB engines will gain a lot from the SSD. These are patterns when a lot of random IO is required. It is usually index access. So there are no specially built RDBMS engines for the SSDs, while IMHO all of them will utilize random IO in some cases.

David Gruzman
+1  A: 

Valentina database can be on of the best choice. And not only for SSDs. :)

Valentina performs queries faster of SqlLite (of course), mySQL, MS, Oracle, FireBird, Postgre at rate x10-100 usually.

It is not huge, engine only 4-10MB (depends on libs you are using). and it create quite compact dbs and indexes.

I need one INSERT query per few second and statistic processing of inserted lines per 24 hour.

for example in one bench with table with 17 fields of all types, on modern MAC INTEL book, 100K inserts go in 3-4 seconds only.

Ruslan