Comments for ScriptBits, Little Bits Of Code You Cannot Live Without http://www.scriptbits.net What are you coding? Tue, 14 Jul 2009 16:13:23 -0500 http://wordpress.org/?v=2.8 hourly 1 Comment on MySQL Federated Engine Allows For Multi-Master Replication by Patrick Kennedy http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-16 Patrick Kennedy Tue, 14 Jul 2009 16:13:23 +0000 http://www.scriptbits.net/?p=5#comment-16 For anybody stumbling across this thread, Justin's MySql blog has a ton of excellent and in depth information on it. For anybody stumbling across this thread, Justin's MySql blog has a ton of excellent and in depth information on it.

]]> Comment on MySQL Federated Engine Allows For Multi-Master Replication by Patrick Kennedy http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-15 Patrick Kennedy Mon, 13 Jul 2009 23:37:52 +0000 http://www.scriptbits.net/?p=5#comment-15 This wasn't designed with a sharded environment in mind. For performance issues we ended up with multiple masters, each with their own set of read only slaves, running unrelated databases with nothing in common but for they are each accessed by our inhouse applications. For reporting issues, our business objects team needed a solution where they could query across these multiple databases. This does that. It also will be able to lend itself useful as we begin to move some apps/services to a cloud environment and have others still in a datacenter....once again aggregating our multiple db's into one place. Table creation and table alters hurt this model. I have a script that monitors this and intervenes to run the proper sql on the proper place. I'll be posting that one in a bit. This wasn't designed with a sharded environment in mind. For performance issues we ended up with multiple masters, each with their own set of read only slaves, running unrelated databases with nothing in common but for they are each accessed by our inhouse applications. For reporting issues, our business objects team needed a solution where they could query across these multiple databases. This does that. It also will be able to lend itself useful as we begin to move some apps/services to a cloud environment and have others still in a datacenter....once again aggregating our multiple db's into one place.

Table creation and table alters hurt this model. I have a script that monitors this and intervenes to run the proper sql on the proper place. I'll be posting that one in a bit.

]]> Comment on MySQL Federated Engine Allows For Multi-Master Replication by Justin Swanhart http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-14 Justin Swanhart Mon, 13 Jul 2009 22:21:44 +0000 http://www.scriptbits.net/?p=5#comment-14 Also replication of table creation is problematic. Though that is the same issue as using an intermediary slave server with the blackhole storage engine for relay log filtering. Also replication of table creation is problematic. Though that is the same issue as using an intermediary slave server with the blackhole storage engine for relay log filtering.

]]> Comment on MySQL Federated Engine Allows For Multi-Master Replication by Justin Swanhart http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-13 Justin Swanhart Mon, 13 Jul 2009 22:18:22 +0000 http://www.scriptbits.net/?p=5#comment-13 My question wasn't about master failover. Normally, the only time you have more than one set of masters is if you have a sharded environment. If you choose to shard your data, either you only use a hash algorithm to decide what partition information is found on, or you use a "hive node", a dedicated master that maps partition keys to shards. The latter is a more scalable model, because it avoids hot spots which invariably develop in a hash partitioned shard. If you have a "hive mind", you can move a partition key to an arbitrary node to balance the load, but it breaks the replication model. My question wasn't about master failover. Normally, the only time you have more than one set of masters is if you have a sharded environment. If you choose to shard your data, either you only use a hash algorithm to decide what partition information is found on, or you use a "hive node", a dedicated master that maps partition keys to shards. The latter is a more scalable model, because it avoids hot spots which invariably develop in a hash partitioned shard.

If you have a "hive mind", you can move a partition key to an arbitrary node to balance the load, but it breaks the replication model.

]]> Comment on MySQL Federated Engine Allows For Multi-Master Replication by Patrick Kennedy http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-12 Patrick Kennedy Mon, 13 Jul 2009 21:44:25 +0000 http://www.scriptbits.net/?p=5#comment-12 You wouldn't in this model. This not for failover master, this is for a database server to have two (or more) separate replication chains written to it. You wouldn't in this model. This not for failover master, this is for a database server to have two (or more) separate replication chains written to it.

]]> Comment on MySQL Federated Engine Allows For Multi-Master Replication by Justin Swanhart http://www.scriptbits.net/2009/07/mysql-federated-engine-allows-for-multi-master-replication/comment-page-1/#comment-11 Justin Swanhart Mon, 13 Jul 2009 21:17:33 +0000 http://www.scriptbits.net/?p=5#comment-11 How do you move data between master nodes? How do you move data between master nodes?

]]> Comment on SED Tutorial and Examples by AWK Scripts and Examples | ScriptBits, Little Bits Of Code You Cannot Live Without http://www.scriptbits.net/2009/07/sed-one-liners/comment-page-1/#comment-10 AWK Scripts and Examples | ScriptBits, Little Bits Of Code You Cannot Live Without Mon, 13 Jul 2009 13:57:07 +0000 http://www.scriptbits.net/2009/07/sed-one-liners/#comment-10 [...] ScriptBits, Little Bits Of Code You Cannot Live Without What are you coding? « SED Tutorial and Examples [...] [...] ScriptBits, Little Bits Of Code You Cannot Live Without What are you coding? « SED Tutorial and Examples [...]

]]> Comment on How to identify what processes are generating IO Wait load. by ddctomc http://www.scriptbits.net/2009/07/how-to-identify-what-processes-are-generating-io-wait-load/comment-page-1/#comment-7 ddctomc Sat, 11 Jul 2009 13:58:03 +0000 http://www.scriptbits.net/2009/07/how-to-identify-what-processes-are-generating-io-wait-load/#comment-7 No both distros (and i think most/all others) are on a 2.6 kernel now. Its just the difference in the systat package version. Hopefully RH/CentOS will incorporate the newer version in a future release. No both distros (and i think most/all others) are on a 2.6 kernel now. Its just the difference in the systat package version. Hopefully RH/CentOS will incorporate the newer version in a future release.

]]> Comment on Create Automated LVM Snapshots for MySQL Backup by Jonas http://www.scriptbits.net/2009/07/create-automated-lvm-snapshots-for-mysql-backup/comment-page-1/#comment-6 Jonas Thu, 09 Jul 2009 21:09:53 +0000 http://www.scriptbits.net/2009/07/create-automated-lvm-snapshots-for-mysql-backup/#comment-6 Do you know mylvmbackup? - http://www.lenzg.net/mylvmbackup/ It supports hooks,tar and rsync for backups and it's using lvm snapshots. Do you know mylvmbackup? - http://www.lenzg.net/mylvmbackup/ It supports hooks,tar and rsync for backups and it's using lvm snapshots.

]]> Comment on How to identify what processes are generating IO Wait load. by Ted Silas http://www.scriptbits.net/2009/07/how-to-identify-what-processes-are-generating-io-wait-load/comment-page-1/#comment-5 Ted Silas Thu, 09 Jul 2009 19:02:20 +0000 http://www.scriptbits.net/2009/07/how-to-identify-what-processes-are-generating-io-wait-load/#comment-5 Kernel 2.6.20 and above maybe? Kernel 2.6.20 and above maybe?

]]>