Hi Steven,

> Can some of you please let me know what type of hardware specs you're using 
> in your data center for your core encoding server(s)? We're asking for 4 
> testing VMs in our data center to be spooled up, and I need to know what 
> virtual specs to ask for.

as I am sure you know there is no definite answer to your question as it 
depends on the amount of data you want to process, the time it needs to be 
processed in, the number of recordings you want to process in parallel etc. 

Generally it can be said that the overall number of cpu cores available in your 
system will determine the amount of parallel work that can be done. Here at ETH 
Zurich we use 4 virtualized RedHat systems with one core each for our test 
cluster, which basically yields a throughput of 0.7 (hours of SD recording 
material : processing time).

Given that you distinguish between the common server types in a distributed 
setup (admin, worker, distribution, engage), this is what can be said about 
each of them:

- admin: optimize for memory, as it hosts search indices to serve the uis.
- worker: needs as much cpu as possible, while memory can't hurt.
- distribution: refer to requirements specified by your distribution server 
(e.g. apache httpd, red5 streaming server etc)

Most importantly, you need to make sure to use shared storage amongst the 
servers. Otherwise you will lose tons of time copying large media files around.

Tobias
_______________________________________________
Matterhorn-users mailing list
[email protected]
http://lists.opencastproject.org/mailman/listinfo/matterhorn-users

Reply via email to