I recently wrote about tying together disparate management tools, and I made a few suggestions in this newsletter to a reader who was dealing with that issue. Now other readers have suggested their ideas, and I’d like to pass those along.

One reader wrote:

“I realize that there are many possible answers. We fought the same battle for years until implementing ‘SuperAgent’ sold by Fluke, really a NetQoS product. I manage servers and networks for a large hospital and we were always getting, ‘xxx is slow; it looks like a network problem.’ Finding out ‘what’s wrong’ complaints and spinning our wheels like your reader trying to prove what the problem was or was not. Pulling together all this diagnostic data effectively was tedious and time consuming. Most non-technical people didn’t understand all the results anyway.”

He goes on to say:

“After implementing SuperAgent, we don’t ‘pull together’ anything at all, and we now can e-mail snapshot graphs and metrics of network response time, server response time, data transfer time and retransmission delay time within minutes. It stopped our finger-pointing overnight and gave us an opportunity to offer objective measurable performance statistics that we could all review and discuss together. It became extremely easy to show what was operating normally and what was struggling under load so the right resources could be focused on the problem. Of course, you can do benchmarking and proactive monitoring with alerting also.”

Another reader shared another possible solution:

“Tying together all of those tools or data by brining in a [manager of managers] seems like a messy and most likely unproductive process. It is kind of like saying that just because I have all the sensors in an automobile and I can see them all at once I will know the answer. I think there would be times all the sensors are green and other times a lot can be red - neither of which solves your problem. There is some promise in upstream and downstream knowledge but trying to get all that to work together when the tools are disparate… seems a bit hopeless to me. Do you see some promise in the idea of starting with an application-level (top-down) triage of the problem and them working downward? Start with the end user experience as the true indicator, then see if it is slow on the network or the data center, then dig into which tier was slow. At least this way you are narrowing the problem down.”

Denise Dubie is senior editor with Network World.