Great points there... In my case, I started with VM in VM/370 Release 6 BSE back mid to late '70s. My "gut" on VM is pretty good, until you really push something to it's limits.
Linux, I'm not sure about. Not enought experience to know if I can fly by the seat of my pants. But the multiuser applications (Websphere and Oracle) are where the resources will be used. And that needs monitoring. And then, there comes the....well I can do it, but then I can't offload it to others. Tom Duerbusch THD Consulting >>> <[EMAIL PROTECTED]> 02/01/05 1:47 PM >>> <snip> But if VM says a machine is using 50%.... And in that machine, TOP says some process is the major user... Then, you look at that process. </snip> That works when only one guest machine is going goofy at a time, but can get far more complex to figure out when multiple guests are busy simultaneously. Just as an example we had a guest get really slow just the other day. The other guests were fine. So we spent a bit of time looking at top, vmstat, etc. but when we went into Velocity it quickly became apparent that the problem was actually on a completely different guest which had gone haywire. Could we have figured it out without the tool? Sure, but we would likely have had to go into each of the other guests to track down the culprit.. <snip> But if the fancy tools then say, this process is using xx service, to reduce the cost of this service, tune yy, then it's a good deal. <snip> Velocity does have some report/recommendation tools.. IMHO it doesn't get quite that specific though and most of the suggestions are VM centric (as one might expect given it's heritage and in our case a good thing). Of course without a complete set of tools to look at if from both ends of the pipe its a problem either way. If you go the way of the oracle tools, how long do you keep tuning oracle before you begin to think you are having a VM issue and vice versa. I guess in the end, maybe the tool you want to look at is the one you are least comfortable turning the knobs yourself. In our case we had absolutley zero VM experience in house when we started doing this, wheras we had quite a bit of Unix/Linux experience. So we opted to have something with blinking red lights on the VM side so at least it would tell us where things might be broken. The fact that we could do that and associate the events from the netsnmp Linux data was just a bonus. My $.02 YMMV.. ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
