As you might know, I am doing a lot of crazy things with Amazons EC2 Servers + their Storage (S3) … a lot of thinking goes into cheap scaleability for startups and in my case mostly scaling streaming video infrastructure.
The main problem so far working with ec2 instances was, that you couldn´t store data reliable on the running server, but had to use amazons s3 storage (or another server) instead. This results to scripting tons of half reliable backup + restore scripts … and all goes bad if you have to work with a relational database (eg mysql) … which made me go crazy concerning reliability of data.
Today amazon launched a service called elastic block store, which seems to solve all there problems. In the end block store is a mountable device which you can hang in any ec2 instance. The devices can hold up to 1TB of data and standart customers can create up to 20 devices …that sums up to 20TB of persistant storage.
I started testing today and i had some timeouts trying to sync s3 + persistant block storage using a ec2 instance – but in the end everything worked fine for me … next stop will be running mysql databases on this storage.
Unfortunately you can´t mount this persistant storage read only on more than one machine. So I´ve got to find another solution to this … but all in all this made my day