Scalr Server and EC2 Instance Sizing

[Jay Farschman]

QUESTION

Has anyone fiddled around with the smallest possible instance size to run Scalr in EC2? What about breaking it into MySQL and everything else?

DETAIL

For our Dev and QA we use the Community Edition all-in-one sized as an m3.medium.  With 175 instances running inside it, I'm getting a CPU graph that looks like the attached image.  One interesting data point is on 7/7 when the server wasn't properly operating ZeroMQ. CPU usage drops to around 30%. This server will be growing to over 1,000 nodes by the end of the year and it looks like I need to start breaking this server into multiple parts.  Nice notes here on a distributed architecture.

I'm asking about this because we have built a second server to manage production infrastructure. The number of instances controlled by this second server will be under 50.  Curious how small I can go?

t2.medium

t2.small?

Two t2.small instances, one for MySQL and one for everything else?

[Marc O'Brien]

Hi Jay,

4gb ram is our minimum requirement for a Scalr install.  2x m3.medium for your larger deployment (1000 instances) and 1x m3.medium for the smaller (50 instances) would be sufficient.  Please note, you will want to use m3 vs t2 instances as burstable instances can lead to poor performance if you have long running workloads.

Let us know if you have any questions.

Many thanks,
Wm. Marc O'Brien
Scalr Technical Support

[Jay Farschman]

Thanks Marc,

That's solid advice.  When you run 2 X M3.Medium, are you breaking it down where MySQL is on one node, and everything else if on the other?

And a side note to anyone who reads this in the future. I tried a t2.small and it worked, but I don't think I'd use it for more than a proof of concept with fewer than 12 nodes.

[Marc O'Brien]

Hi Jay,

Yes that would be with MySQL on one node and the everything else on the other.

Cheers,

Wm. Marc O'Brien
Scalr Technical Support