[Spread-users] Real World experience with large spread networks?

[John Schultz]

Remember that each daemon can support hundreds of clients.  So one
question to answer is whether or not you really need a Spread daemon
on every one of your machines or whether you can design your system
to have clients on lots of machines and daemons on a (much) smaller
subset of machines?

The ring protocol, as it is, has inherent scalability limitations,
which is why there are hard limits, etc.

On the spread.conf file distribution question, using NFS and rsync
are common solutions.  Another alternative would be to distribute it
through Spread itself (although this won't help with the machines you
will be adding).

Cheers!
John

---
John Lane Schultz
Spread Concepts LLC
Phn: 443 838 2200 
Fax: 301 560 8875

Tuesday, June 16, 2009, 12:18:05 PM, you wrote:

> I'm working on a project which will be setting up a spread network
> across a large number of nodes: perhaps 50 to start with and then
> increasing up into the hundreds...

> Reading the Spread User Guide (Section 2.4.4), I've found 2 interesting facts:

> a) There's a hard limit of 128 machines in a configuration

> b) The protocol hasn't been tested with more than 50-60 daemons.

> Does anyone have any real world experience they'd be willing to share
> about running spread over a large number of machines? Are these limits
> still current?

> Also, given that each machine must have the same spread.conf file (if
> I understand it correctly), then how do people solve the configuration
> distribution problem when they add a new machine to the network?

> Rachel

> _______________________________________________
> Spread-users mailing list
> Spread-users at lists.spread.org
> http://lists.spread.org/mailman/listinfo/spread-users