QoE conversations and grouping

I have just tried QoE for a while but I can't realy get my head around how you have designed this functionality.
In a larger enviroment it is pointless to get the average of a specific application. I have warnings on SSL but there are like 100 servers running SSL! wheres the problem? a specific datacenter? a specific system? what servers?

1. I would realy like to be able to Group Servers from an System perspective to see what System (like 'payroll system X) to see that there is slow responses and then drill down to what server(s) reported the slow respons

2. I would like the same (and a combination) of a topology perspective (sites/datacenters), this is to see that CIFS have slow network responcetime in London Office.

Secondly

Why dont you use this exelent agent to get also the conversation information? It would be realy something to see what clients use a services on a specific server. Some one may say that 'this is why you have netflow (NTA)'

sure, but there you dont have the response-time.

What i realy would like to see is:

a) What 'System' have problems with slow responstimes

b) What application and server have the problem (like: SQL is slow on server XXX)

c) what clients have the problem? many?, few? all?

d) If the slow response is from the network, on what sites are theese clients located (site = subnet(s))

One of the usecases for us would be to plan where to locate our file servers and to see that all sites have good enough respons-times.

This would support server consolidation and minimize the spawl of local servers.

I would realy like to get some feedback on this to see if QoE could be a solution for us or if I need to keep searching for a solution.

cheers!

Find more posts tagged with

qoe

responstime

conversations

Status: None

Comments

alanaround

I agree with many of your points. I have just deployed QoE and I've never seen a dashboard provide so much data that means so little. I have CIFS that is in the seconds, and 'red', but I have no idea what specific connections it is referencing. With the amount of data I am providing the sensor (raw SPAN port with large amount of data) what we actually get out the other end is poor.

The concept is good, but it doesn't really help me drill down to resolve or troubleshoot where the actual issues are. I was hoping to see an actual conversation window with source, destination and then the top XX poor conversations or whatever, or at least by subnet, so I could start focusing on a poorly performing remote location.

familyofcrowes

I had high hopes as well, but I think this is just the first version of QOE and it will get added features with upcoming versions. The data is there like you said, so I'm sure we will see improvements over time. At least thats what I'm telling my management :-)

alanaround

I guess thinking about it, if we purchased the desktop/endpoint module (can't remember the name) and therefore had all of our desktops as nodes in NPM, then QoE would report on them too, but still not as conversations, but at least we'd be able to say that this endpoint or group of endpoints (constituting a remote site) are all having issues with CIFS or whatever... still, i'm not really sure i want to import all of my desktops into NPM..