A Client's View Of App Monitoring
For years I have been a huge advocate of monitoring applications and services.Unfortunately, my methods of choice for monitoring have been a hodgepodge of customprograms, off-the shelf applications and a lot of luck. My approach has been pretty muchthe same as every other "management expert" in the field: to look from theinside out. In many instances, that means looking at the server, not the client. Forexample, we developed an e-mail monitoring application a few years ago to monitor acc:Mail implementation.
The application in-volved monitoring a text file for errors on DOS PCs, file systemconnections between those PCs and Windows NT servers, as well as gateway applications onNT and UNIX systems. Each of these custom applications wrote its data to a central logfile on a UNIX machine. We then generated SNMP traps based on what events were written tothe log files. This long involved process finally provided us with application monitoringevents in our OpenView event browser. But, that's not the end of the story.
We started finding ourselves building little monitoring systems to verify and keep thee-mail monitoring system running. Don't get me wrong. The system actually was a work ofart that was extremely reliable, as long as it had plenty of care and feeding. I havealways been an advocate of SNMP management but few vendors or software developers havebeen willing to instrument SNMP in applications. There have been exceptions but nowidespread implementations yet. Look for SNMPv3 and it's inherent security to change someof that. However, even a new simpler solution still involves looking from the inside outor simply looking and the server.
What's the alternative? Looking from the outside in? No, you have to look atapplication monitoring from the client perspective. International Network Service (INS)has a product called Vital Suite (www.ins.com/software /vitalsuite) that performs thisunique client perspective monitoring. It's based on an agent called VitalAgent IT. ThisAgent collects all relevant system information as well as performance data and applicationtransaction data from the client. The next piece of the puzzle is the VitalAnalysis tool.
Vital analysis periodically collects all the agent information and correlates ittogether. It represents all this data it in a rather unique Web-based interface. Both theagent and analysis products have excellent interfaces that can actually be quite addictingto use. The next puzzle piece is the VitalHelp which is the real-time event manager.Besides all the cool technology working between these products, the best part is thegateways that exist to move data from the VitalHelp product into your existing managementand help desk systems.
Vital Suite has 3 gateways available SNMP, SNTP and Remedy. I am also told, for thereal diehard developers out there, that a SQL interface is available to the VitalHelpdatabase. So, we are monitoring applications and their response time from a truecustomer's perspective. How can you go wrong?
PROMISES AND LIES
Now some products and companies such as Compuware have claimed to do this for years byevaluating network traffic and performing measurements. The problem with Compuware andothers has been two fold: First, monitoring is still performed from a central location, soyou miss the "true" client perspective. The second problem is getting thatmuch-needed application uptime and downtime data into a central management system such asHP OpenView.
Don't misunderstand me. I'm not advocating dumping all SNMP network, system andapplication management systems for tools such as Vital Suite. What I am proposing issimplifying some of our current specialty management systems and complementing them with"client perspective" monitoring.
-- Charles Hebert is President of Southernview Technologies, Inc. He canbe reached at firstname.lastname@example.org.