Reply to Message

Real performance monitoring ? look out for tractors!

Make IT perform Updated - 17th Nov 2008

Body : While we can agree that understanding the detail of transactions can be the key to spotting application related problems, both front and back end, I cannot agree that this will enable an organisation to ,as you say, ?... grasp the intricate web of connectivity between those business activities and the IT infrastructure components.?

Measuring specific transactions alone is simply not enough to understand that interconnectivity, it ignores so much else that is going on. Just monitoring key application transactions is like running a road transport operation by measuring the performance parameters of just your super important red vans ? you can tell where they are, and when they got there, and learn when there is a hold up. All useful information, but not really as useful as being able to also spot the tractors, broken down trucks and road restrictions that are causing the hold ups. Or having trending information at your finger tips which will let you understand delays are due to ?rush hour? traffic rather than a red van problem.

For real insight you must look at the whole system picture to give context to the performance metrics of individual applications. You need for example to correlated application performance with trends in SNMP data to understand link utilisation impacts and Netflow data to attribute and investigate application and user usage. Only then can you optimise, diagnose faults and plan your systems effectively.?

Benny Vogels
Fluke Networks