|Performance Analysis and Tuning Simplified|
System Administrator's Page
If you're a System Administator, you never have enough time to do it all. And you probably get stuck trying to explain system problems and possible solutions to a non-technical person (like managers and the ever-present "bean counters"). That's where SarCheck comes in.
SarCheck is a performance analysis tool for most Solaris, AIX, HP-UX and Linux systems. SarCheck is less expensive than just about anything else out there except freeware and we're sure you know that the owner of your company pays attention to price. It enables you to quickly understand the cause of most common performance problems by explaining things in a plain English report that you can show to management. SarCheck can use gnuplot to create graphs that make it even easier to explain what's happening on your system. Here are some pages with moderately geeky content that show some of the graphs produced on UNIX and Linux systems. Now you can finally prove that you need an upgrade and you're not just looking for new toys! Here are links to sample HTML formatted SarCheck reports for Solaris and Linux operating systems.
SarCheck does not link into the kernel, you don't have to reboot after you install it, and it usually installs in under one minute! When you upgrade to the latest version of SarCheck, the new version overlays the old one so you don't even have to de-install. When SarCheck makes recommendations, it explains the reasons for those recommendations so you can decide for yourself if it's telling you the truth. It is incapable of changing things on its own, so you remain in control of your system and will get the credit when SarCheck's recommendations help to make your system fly. It also uses minimal system resources. SarCheck will typically use about one second of CPU time to produce a report and requires less than a megabyte of disk space.
If you've had to wade through your system searching for the cause of a performance problem, you understand the advantage of automating this thankless job. SarCheck will recommend both system tuning and hardware upgrades where appropriate. SarCheck also provides some basic capacity planning data. In many cases SarCheck can tell you how much of an additional workload a system can support, and which system resource is likely to become exhausted first. You'll be the first to know when your system needs tuning or a hardware upgrade, and you'll know early enough that everyone can see that your system is in good hands.
If you take care of lots of systems, you can even run SarCheck every night as a cron job, and mail the output to a central site or post the HTML output on your intranet. Because the output is stdout, you can do almost anything with it. Various filtration options are available to prevent SarCheck from overwhelming you with information if you have it running on 100 different systems, and many of the thresholds used by SarCheck's rule base can be overridden.
All sys admins: Here are a few of the reasons that SarCheck is the way for you to go.
Solaris admins: SarCheck is a product that was designed to run on production systems. Of course you should try it out and make your own decision, but it's a real product which is fully supported by its manufacturer and you can read all about it on the SarCheck for Solaris page. And don't miss our new segmap algorithm, we're still finding new ways to wring more performance out of Solaris.
AIX admins: In addition to looking at sar and ps data, SarCheck for AIX analyzes numperm data and graphs it if you're running gnuplot. If you want to see more information about maxclient, maxperm, minperm, and numperm, visit the SarCheck for AIX page.
HP-UX admins: Back in 1999, the Interex HP-UX System Performance Ballot identified the need for better performance tuning guidelines as the #1 issue. We've had a solution to this problem for years and our solution keeps getting better.
Linux admins: SarCheck for Linux is here and more information can be found on the SarCheck for Linux page. SarCheck for Linux gets its data from the /proc filesystem. Even though there's a really good sar utility out there, we get better granularity by going straight to /proc.