I've been doing a lot of research into scaling issues on Amazon EC2, especially growing databases. I looked into how I can make the database more solid and sound (I use PostgreSQL). Many sources say using RAID0 with streaming replication and WAL archiving is pretty fault tolerant.

All the postresql data & tables will be stored on the EBS volumes that have RAID0. Many sources claimed that using an XFS system would be ideal. Is it possible to add more EBS volumes to this RAID0 array if I ran out of space while keeping the database online? If so, is there some sort of recipe or source I can follow?


Assuming you're using Linux's md software RAID I'm pretty sure can hot-add volumes to a RAID 0 using the --grow command, but I would recommend testing this to verify that I'm right and that it's supported by your md version.

If you're planning on using real-time replication anyway and you're confident you can run a good, reliable replication and backup system you might as well use the instance store rather than EBS, especially if you can justify a High I/O instance. Remember that if you terminate an instance store based instance or it crashes you will lose the data on that instance permanently, so backups and replication to other AZ's / regions are vital.

Personally I prefer to run DBs on other hosting providers that're cheaper and offer much better I/O performance, but EC2 has some management advantages.

I wrote a post on this a little while ago that may be useful: Improving PostgreSQL performance on EC2.

  • hey! I came across your post a few months back, it was very useful, I'm trying to find a really good tutorial for RAID0 setups – noahandthewhale Apr 19 '13 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.