Comments on: Finding your MySQL High-Availability solution – Replication http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/ Everything about MySQL Performance Thu, 29 Jul 2010 19:06:57 +0000 http://wordpress.org/?v=3581 hourly 1 By: yves http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/comment-page-1/#comment-712436 yves Wed, 20 Jan 2010 21:06:24 +0000 http://www.mysqlperformanceblog.com/?p=1607#comment-712436 Hi sv, the manager process is very light so there is no problem managing more than one pair of servers from a single manager server. Another option is to use your monitoring/alerting server (like Cacti/Nagios) to host the manager process. Hi sv,
the manager process is very light so there is no problem managing more than one pair of servers from a single manager server. Another option is to use your monitoring/alerting server (like Cacti/Nagios) to host the manager process.

]]> By: sv http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/comment-page-1/#comment-712395 sv Wed, 20 Jan 2010 18:13:46 +0000 http://www.mysqlperformanceblog.com/?p=1607#comment-712395 We have test servers, beta servers and evaluation servers along with the production DB server. Will I be able to use just one server as the monitor and group these servers? I mean a single manager should manage master-master DB server for test server. Then another group in the same manager managing another master-master DB pair for the beta server. Will this be possible? I do not want to waste an entire server for just the manager software. Thanks. We have test servers, beta servers and evaluation servers along with the production DB server. Will I be able to use just one server as the monitor and group these servers? I mean a single manager should manage master-master DB server for test server. Then another group in the same manager managing another master-master DB pair for the beta server. Will this be possible? I do not want to waste an entire server for just the manager software.

Thanks.

]]> By: R http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/comment-page-1/#comment-706635 R Thu, 07 Jan 2010 12:39:51 +0000 http://www.mysqlperformanceblog.com/?p=1607#comment-706635 The trick here is to run the alter table on a slave and then, once done, we let the slave catch up with the master using the new table schema, we failover to the slave and repeat the alter table on the other server. Those online schema change are easier when a master to master topology is used. --- Please describe this a bit more. What does "failover to the slave" mean? How will master get synchronized with new slave scheme? The trick here is to run the alter table on a slave and then, once done, we let the slave catch up with the master using the new table schema, we failover to the slave and repeat the alter table on the other server. Those online schema change are easier when a master to master topology is used.

—

Please describe this a bit more. What does “failover to the slave” mean? How will master get synchronized with new slave scheme?

]]> By: yves http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/comment-page-1/#comment-700890 yves Thu, 24 Dec 2009 16:26:06 +0000 http://www.mysqlperformanceblog.com/?p=1607#comment-700890 Hi Mahesh, master-master and master-slave are very similar, master-master is basically 2 master-slave setups. Doing writes will no improve performance since both servers will need to perform the writes and will likely lead to collisions. The goal with the MM setup is to ease the failback. Hi Mahesh,
master-master and master-slave are very similar, master-master is basically 2 master-slave setups. Doing writes will no improve performance since both servers will need to perform the writes and will likely lead to collisions. The goal with the MM setup is to ease the failback.

]]> By: Mahesh http://www.mysqlperformanceblog.com/2009/11/13/finding-your-mysql-high-availability-solution-%e2%80%93-replication/comment-page-1/#comment-699612 Mahesh Tue, 22 Dec 2009 07:05:43 +0000 http://www.mysqlperformanceblog.com/?p=1607#comment-699612 Hi, Thanks for this informative article. I am confused between choosing master-master(MM) & master-slave(MS). What is the difference between MM and MS if in both architectures, writes go only to master and reads can happen from both? Also in MM, why do the two masters replicate each other when writes are happening on only one? Hi,

Thanks for this informative article.

I am confused between choosing master-master(MM) & master-slave(MS).

What is the difference between MM and MS if in both architectures, writes go only to master and reads can happen from both?

Also in MM, why do the two masters replicate each other when writes are happening on only one?

]]>