On 8/6/07, John Stanton <[EMAIL PROTECTED]> wrote: > > > We use a single process server as an Sqlite server. It works well > because it obeys certain constraints: > o Transactions are always short > o It has many users and many Sqlite databases, but each database > does not have a large number of users, or more correctly a large traffic > rate. > o Row scans are avoided on large tables. > > The benefits are the ease of maintenance of multiple databases, each one > being just a file. Transactions complete typically in less than 500uS.
sounds good. > > We use HTTP protocol to access the Sqlite server, and it allocates one > or two threads to each user connection (a browser will try to open two > connections). > > If you have large transactions or large numbers of users sharing a > database, consider using something like PostgreSQL, Oracle or DB/2. > Sqlite is inherently single streamed and that imposes a cap on its > ability to handle many simultaneous connections to a single database. this is a personal project and i really want to use sqlite. thanks for this. i'm a big pg fan. <snip> thank you. ./e -- no sig ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
