Having done this a few times with mysql-to-mysql moves across datacenters theres a couple things I would like to mention.<p>I dont know what Amazon RDS is like replaying binary logs but in mysql 5.0 I faced a bug where temporary table creation statements were being ignored. This was a huge problem from us and at the time could not find a work around. Here is the related mysql bug <a href="http://bugs.mysql.com/bug.php?id=35583" rel="nofollow">http://bugs.mysql.com/bug.php?id=35583</a><p>While this move has successfully worked for the author I hope if you ever do a database move you'll do a test run first. I've experience numerous bugs in mysql replication and replaying binary logs due to temporary tables, duplicate keys due to different auto insert ids and procs/trigs/funcs causing some havoc.
I posted a comment on the blog, but will leave one here as well.<p>In regards to the claim of: "quick instance size scaling with no downtime"<p>I am currently evaluating moving our MySQL server to Amazon's RDS. I liked the sounds of the Multi-AZ RDS as it will greatly decrease downtime. However upon examination I found the following issues:<p>- Changing an instance size results in downtime. Amazon's docs and support say up to 3 minutes.<p>- Any failover can take up to 3 minutes to promote the backup to becoming live.<p>- You will experience downtime during that time<p>- Browsing the support forum, some people complained of it taking more than 3 minutes to failover and in some cases the failover got stuck and they experience a longer downtime until they wrote on the forum and had AWS support manually fix it.<p>So in other words, this is not a silver bullet to making MySQL have 100% uptime. And in fact you will experience "up to 3 minutes" of downtime each week during the maintenance window when a failover will occur (unless they do the failover before the maintenance, which I have not found information on anywhere).
I've been planning my move of a 25gb MySQL db over to RDS.<p>I decided to have a downtime window in the middle of the night, versus doing a live move. I am moving from a dedicated datacenter to AWS all at once, so I'm in a different situation than those moving from an EC2 mysql instance to RDS, which should be a lot easier.<p>I followed the instructions from the RDS instructions, which tell you to break up mysqldump files great than 1GB and use mysqlimport to send them over.<p>I did some benchmarking and found the import time wasn't linear on file size. This may have to do with the fact that I was using the 1.7GB instance (which i'm upgrading to 15gb for the live site).