Thanks - I'm one of the authors of that presentation. The DGTIC provides computing to many gov't agencies. Yeah getting a headline 'bout a computing disaster is one of the worst exposures for a government. Management needed to convinced of the soundess of the z/VM and linux on system z solution - so we needed to show that the risk of injurious outages was minor. The proof of the solution was an early success with DR. One outage over 18 months also has helped! (It happened while I was on vacation too). Also I posted a few weeks back about a lesser publicized benefit of using a performance product - defense. In short we have been able to deflect issues away from us because we can show *EXACTLY* what our z/VM environment was up to. David
________________________________ From: Linux on 390 Port on behalf of barton Sent: Fri 3/16/2007 6:14 PM To: [email protected] Subject: Re: Sizing an HTTP-server guest I saw a great presentation from DGTEC - Quebec Government. One of the drivers for management is to NOT get their face pictured in the paper because of IT disaster. Maybe a call to a local newspaper reporter interested in how the county is being serverd will make a difference? Dave Jones wrote: > I realize that I must sound like a broken record on this topic (or is > that a "broken CD" to today's youngsters? :-), but this is where having > a good Linux and VM performance monitoring and analysis product can > really pay off. Instead of having to do a trial and error approach to > guest sizing, one could look at the memory usage data from the monitor > and be able to make a realistic determination from that. > > In today's complex and mission critical systems, having a good > performance monitor and analysis system running is just as important, > imho, as having the proper security manager and policies in place. No > one would consider running a z/VM system out of the box with having > RACF/VM, ACF2, TopSecret, etc. installed as well....the same should hold > for the performance tools available from IBM and Velocity Software. > > Have a good weekend, too. > > DJ > > Rich Smrcina wrote: > >> A customer has one defined at 96M, no swap. >> >> James Melin wrote: >> >>> Greetings everyone. >>> >>> We're working on separating the HTTP server function from the server >>> on which WebSphere runs (Network deployment configuration, where HTTPD >>> was >>> running on node 1 of the cluster). I know I'll be able to shrink the >>> guest considerably but I'm really not sure how to look at the memory >>> allocation >>> of the process now, under workload, to know how big the guest should >>> be. I'm guessing it will be under 300 MB but I'd like to do better >>> than a guess. >>> >>> We're using the IBMIHS server which is based on Apache. >>> >>> If anyone has any insight, I'd greatly appreciate it. >>> >>> -J ---------------------------------------------------------------------- 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 ---------------------------------------------------------------------- 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
