I don't think anybody has done a thorough analysis on memory requirements
for each node type.  The JVMs do not need much, 512 is probably fine, but
certainly 1GB is sufficient.  If you want to be able to run multiple
concurrent encoding jobs (which run outside the JVM), you should probably
have more memory available to execute the 3rd party tools, such as ffmpeg
and tesseract.

For the admin and engage nodes, you may need a little more than a "standard
webserver", especially if you're thinking in terms of Apache HTTPD.  Both
admin and engage are managing things behind the scenes, so require more
memory than what's needed to serve HTTP requests.  Again, though 1GB for the
JVM should be more than sufficient.

The nice thing about VMs is that you can try different configurations to see
how they impact performance.  I'll be interested to hear how your deployment
goes.

Josh

On Fri, Feb 25, 2011 at 11:54 PM, Adam Moore <[email protected]> wrote:

> I've seen the Deployment scenarios on the matterhorn project page and there
> is no real consistent setup for the servers.    The best I can glean is that
> workers should have the most memory and engage/admin can be standard
> webservers.   Has anyone tested what the minimum memory requirements were
> for them?  We will be using vm's so we would look for the best way to use
> the memory efficiently.
>
> _______________________________________________
> Matterhorn-users mailing list
> [email protected]
> http://lists.opencastproject.org/mailman/listinfo/matterhorn-users
>
_______________________________________________
Matterhorn-users mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn-users

Reply via email to