We are testing the IP SLA Monitor and are running a 30 eval to decide if it is something that can help identify issues on our WAN.
Being new to IP SLA, i have a couple questions.
Roughly here is our topology, we have 3 main datacenters, that all connect via a VPLS (Provider#1) on the core 6509 switch (which is acting as a core router), these sites have a 2nd WAN router that has a MPLS (Provider#2) connection to all remote sites. We are running EIGRP between the VPLS sites, and running BGP across the MPLS WAN to all remote locations.
Loc A router 1 6509 (VPLS), router 2 3845 (MPLS)
Loc B router 1 6509, router 2 3845 (MPLS)
Loc C router 1 6509, router 2 3845 (MPLS)
Loc D router 1 2811 (corp office) that only has a VPLS no MPLS
plus Multiple other remote locations with 2851's (MPLS)
Right now, we are more interested in looking at the traffic across the VPLS because traffic between 2 of the sites is experiencing a high latency that does not appear to be explained by utilization or anything else.
I'm new to IP SLA, so forgive my ignorance.
I have tried to set up a couple different types of operations using the wizard to look at the links in different ways, but it appears that i'm being limited to whatever is configured first.
For example, I set up using the mesh network, nodes Ar1, Br1, Cr1, Dr1 looking at UDP jitter. I then tried to do the same thing looking at ICMP Path Jitter, but this fails, is this because I already have operations set up on each router? If i delete what I already have configured, then it appears that I can create the operations that fail. For example, first I set up the operations as "tcp connect", then tried to create the same sort of thing using UDP jitter, the UDP jitter operations failed with a message "Operation setup on the node failed. Invalid input detected at '^' marker. " however, it doesn't show what exactly that input was, or where it failed. By deleting the Tcp connect operations, I could add the UDP Jitter operations. This was now successful. I really wanted to also set up ICMP Path Jitter, so I could compare and contrast, and see which provided more information. But it appears I can only choose one operation per router?
which type of operation would best be used to provide what i'm looking for?
Also, in looking at the wizard, it implies that using ICMP path jitter it looks at hop-by-hop, but how does it identify the hops? For example if I want to watch the traffic from another node at loc A going to a node at loc C, I want to start with a node at A, first hop should be Ar1, 2nd should be Cr1, then nodeC. All the nodes are in Orion, and I have rw & ro strings for each of the nodes. I would think the wizard would be able to determine the hops, but it doesn't appear to do that correctly, and looks like it will say that NodeA directly connects to NodeC even when there are at least 2 routers in the middle. Am I missing something in how it is supposed to discover the hops? In our test eval I can delete and re-add nodes, and let it auto-discover if that would help. I won't be able to do that with our production Orion NPM if we actually decide to purchase this product, we have over 2500 nodes, I can't delete and re-create nodes just to get the SLA working right. Is there a way to force it to identify the hops?
Is there a guide I can use that would help me understand the best way to set this all up?
Any help would be greatly appreciated!!!