We currently have standalone URS deployment with hot standby (7.6, handling inbound/outbound voice interactions). But with increasing load we are considering introducing load distribution with among multiple URSs.
It seems there are two most common ways to load balance, using-
A. LDS with multiple URS
B. Partitioning the route points across instances of URS (limit monitoring by event_arrive option)
I am trying to understand the pros and cons of each approach and curious to know if you know about any issues with any of these approaches or any suggestion about things I should consider.
We did consider restricting with the event-arrive, but it seemed very confusing, hard to maintain, etc - and would also be difficult to balance traffic.
LDS was useful when URS performances were poor; Genesys discontinued LDS, the last version released was 7.2.
I have seen URS working with very complex strategies while handling up to 18 caps, what’s the traffic peak you are experiencing?
I am seeing problem with URS (hosted on IBM P5) traffic reaching at 12 caps with EWT of approx 30 minutes(It just cannot handle the volume). When I spoke to Genesys, they say a single instance is good for handling upto 6 caps.
If EWT is so high, my opinion is that the issue is about the queues being so huge rather than the traffic load; of course the two things are connected, but my point is that I’d try to look at whether you can optimize something:
are you using skill based routing?
how big are your agent groups?
do you have overflow queues? any thresholds which limit the max time a call can stay in queue?
and
[li]what’s the URS CPU usage?
is the CPU usage of the other processes running on the same URS box fine?