Re: SELECT queries on replicate DB server
Thanks, that is a really good answer. Raises a bunch more questions but they're good ones. Jim Grill wrote: My question is: Why would they deem it necessary to use yet a third server? Could it be because the main server and the main slave are constantly being updated and they wouldn't want to overload the main slave(which is not on as high a horsepower of a box I know for a fact). Could it be because maybe the subset of tables that they put on the third server are relatively more stable and hence there arent so many writethroughs so it can handle the complex selects better. All theories gladly accepted... I'm not too sure about the third server either, but I do have an idea. It wouldn't make much sense if the third server had different data on it. That would tend to make things difficult to keep up to date - or maybe not. It might be a slave that they only connect to the master every so often. It's very common to have applications that write to one server and read from a slave server. Sometimes many slave servers since there are typically way more reads than writes. Perhaps they use the third server so that if the master or slave servers die there will always be a spare server for reads. As far as any difference in the tables on the third server... Since it is doing selects only you can start a slave server with a few options to speed things up like: --skip-innodb, --skip-bdb, --low-priority-updates, and --delay-key-write=ALL which will force the server to use non-transactional MyIsam tables for better performance. It's really tough to speculate. Every system administrator would probably do it a different way. Jim Grill -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
SELECT queries on replicate DB server
We have a main db server that has a pretty big db and it sometimes gets overloaded. We also have a replicated slave server... every update to the main server writes through to it and it is the one that is backed up to avoid interruption service for the website. There are certain sections of the website where all the queries are select only. So what was done was to replicate the affected tables on yet a third server. Now we're set up so that when that block of code is invoked, it connects the third db server to take the load off the main one, knowing ahead of time that there won't be any updates to it, everyone is happy tra la la la life goes on. My question is: Why would they deem it necessary to use yet a third server? Could it be because the main server and the main slave are constantly being updated and they wouldn't want to overload the main slave(which is not on as high a horsepower of a box I know for a fact). Could it be because maybe the subset of tables that they put on the third server are relatively more stable and hence there arent so many writethroughs so it can handle the complex selects better. All theories gladly accepted... -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
Re: SELECT queries on replicate DB server
My question is: Why would they deem it necessary to use yet a third server? Could it be because the main server and the main slave are constantly being updated and they wouldn't want to overload the main slave(which is not on as high a horsepower of a box I know for a fact). Could it be because maybe the subset of tables that they put on the third server are relatively more stable and hence there arent so many writethroughs so it can handle the complex selects better. All theories gladly accepted... I'm not too sure about the third server either, but I do have an idea. It wouldn't make much sense if the third server had different data on it. That would tend to make things difficult to keep up to date - or maybe not. It might be a slave that they only connect to the master every so often. It's very common to have applications that write to one server and read from a slave server. Sometimes many slave servers since there are typically way more reads than writes. Perhaps they use the third server so that if the master or slave servers die there will always be a spare server for reads. As far as any difference in the tables on the third server... Since it is doing selects only you can start a slave server with a few options to speed things up like: --skip-innodb, --skip-bdb, --low-priority-updates, and --delay-key-write=ALL which will force the server to use non-transactional MyIsam tables for better performance. It's really tough to speculate. Every system administrator would probably do it a different way. Jim Grill -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]