On 4 Mar 2014, at 5:06am, romtek <[email protected]> wrote: > If that were true, then I wouldn't be getting a very large speed-up when > enveloping write ops in an explicit transaction, would I?
Sorry, I can't tell. It depends on how the virtual storage mechanism works. But if your performance characteristics when using a third-party server resemble the performance of using a :memory: database, then they're using virtualised main storage. If you're finding that when using your own computer, you may find that your main storage is actually a hybrid drive which uses a large solid state cache. Virtual storage is what hosting companies do these days: it uses less power, the servers take up less room, and the lack of moving parts means less failure. Unless your hosting company guarantees ACID behaviour and minimal loss on power failure, which I don't think I've seen in standard cheap hosting contracts. The take-away messages from all of this, if I may be so bold, are (A) Your use of transactions in SQL should reflect which database modifications go together: which ones should fail if they can't all be done. Performance issues are secondary. (B) Your program should run "fast enough", not "as fast as possible". Simon. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
