[Spread-users] performance again
david.turland at shazamteam.com
Mon Aug 5 09:48:30 EDT 2002
I am rapidly approaching the situation where our Spread managed cluster
will be expanded to 60 nodes with
Spread being used, ideally, for both reliability and messaging.
Messaging will be order (100 x 6k messages) / second, comprising,
typically, pairs of (broadcast to worker nodes, one unicast to master node)
I intend for each node to be running a spread daemon.
I imagine there will be a scalability problem with Spread at this number
>3.16.2 has no problem to support 100 receivers but not if each of them
>is running on a different machine. For that you need a smarter design.
Can somebody summarise the issues affecting scalability ,and what of the
following could increase the chances, if any, of Spread supporting the
1- no of nodes running spread daemon
2- message rate , bandwidth
3- reducing the message 'reliability' type
4- moving the unicast to a standard socket call
5- moving the multicast to say, LGMP,LGCP,RMT.. i.e. having Spread just
maintain the reliability
6- A smarter design, or a more scalable version.
What are the issues with 'A smarter design, or a more scalable version'?
Is this design
- available from the commercial arm of Spread.
If available, is it a panacaea or an improvement
so many questions......
Thanks in anticipation,
More information about the Spread-users