Creating meshed operations for 30+ sites

Hi,

there is a reason why we use limitation of 18 nodes(306 operations) for fully meshed configuration that you can create at one time. When configured properly, IP SLA operations have a minimal impact on your overall network health. Problems can arise when configurations force operations to be tested too frequently, or when too many. In your case for 30 devices we have 30*29/2 = 435 links where total operations = 435*2*4=3480 operations.

In case of high number of IP SLA operations you could run into following problems

• Overall network performance will start to degrade
• thousands test results stored every five minutes can create a large database affecting other Orion services
• thousands of IP SLA operations could add a significant burden to the SNMP poller.
• Chance that most of the historical results will never be examined due to the large number of results to filter
• IPSLAManager wizzard may run into timeout when you would try to add more than 18 devices in full meshed configuration

I'd recommend to avoid overlapping tests. But you can also create fully meshed operations on two chunks (15 and 15).

We have 33 sites fully meshed with MPLS.  I manually created the fully meshed IPSLAs 32 from each site for a total of 1056 operations (32*33).  It does not seem to be causing any problems on our networks using the default timers of 300sec.  Just seems like limiting it to 18 is a little restrictive.    To calculate the number of operations it should be (# of sites -1) * (# of sites).

To reduce the load caused by IPSLA we are distributing the operations over that 300 second window so only a few operations are running about every 10 seconds.  Rather than all the operations running at a single time.

By using the group scheduler command to start the IP SLA operations it drasticaly reduces the load on the network to something like 16-32Kbps every 10 seconds in our case using 729a and 1056 operations.  Rather than 256-512Kbps every 300 seconds and yes I agree that is a little much for our network of DS1 to handle at once.

Example.
ip sla group schedule 1 1-36 schedule-period 300 start-time now

If the IPSLA manager created the schedule using this it would help reduce the load on a larger network and would not be a problem.

Hi JustinY,

thanks for your improval hint. It is definitively valid request (as you also mentioned on this thread: