Despite IBM Power Surge, MPG Keeps Performance Tool Relevant
October 21, 2014 Alex Woodie
The folks at Midrange Performance Group have a well-deserved reputation in the IBM i community as performance experts. When it comes to capacity planning exercises and sizing new boxes, its Performance Navigator product is considered top-notch. But how does the company keep PerfNav relevant when the majority of IBM i customers’ workloads can comfortably fit on a relatively small machine? Here, it’s the little things that count.
The huge amount of raw performance that IBM delivered with the Power7 servers (and is continuing to deliver with early Power8 servers) is a double-edge sword for the community as a whole. On the one hand, customers benefit because they simply get more bang for the buck. But it’s made things challenging for hardware resellers, who are selling fewer numbers of more powerful boxes, and software vendors, who must cope with customers shifting into lower (i.e. less expensive) software tier groups.
The advent of relatively cheap Power Systems servers with huge CPW ratings is also impacting capacity sizing exercises, according to Jim Young, vice president of sales at MPG. “…[A]lmost all customers can be accommodated with a 720 or equivalent, which is just a function of Moore’s Law,” he says. “From a sales perspective, it’s made things more difficult because that issue does arise.”
Thanks to IBM’s big investments in the Power processor, today’s entry-level Power Systems servers have more processing oomph than the biggest iSeries and System i servers from the past. While new Java and PHP applications can feast on these resources, the bulk of IBM i workloads continue to be RPG, which is a complete miser with resources. (IBM may have been better off emulating the OS and program bloat that accompanied the last decade of Windows-on-Intel development. But, alas, Big Blue is too beholden to customers.)
There’s pushback from folks who never had a sweet set of performance tools like PerfNav, Young says. “You take a guy on the street and he just bought a 720–he can make a valid case that he doesn’t need this, that it’s more than he’ll ever need. Sometimes that’s true.”
But what model of server you buy isn’t the whole equation. You need to properly configure and outfit the server with the appropriate amount of DASD and memory. There is the ever-present disk-arm sizing question, which some business partners continue to ignore, to Young’s utter amazement. And then there’s the ever-increasing level of virtualization. All of these factor into the continued relevancy of PerfNav.
“As things get a lot more sophisticated in terms of virtualization and LPARs and VIOS . . . the job of understanding what in the world is going on underneath that operating system is getting steadily more difficult,” Young says. “The guys who committed to running their business on this computer–whether the Model 720 or up the product line– they all need a good bunch of the functionality we offer. I don’t see the complexity around virtualization getting lighter. It’s going to get heavier.”
MPG is aiming to keep these customers happy with the latest release of PerfNav, which the company unveiled two weeks ago. With PerfNav version 17, the company is supporting the latest Power8 servers from IBM. That ensures customers will be able to conduct capacity planning exercises, and forecast how their production workloads would fit onto any possible configuration of IBM Power server going back over a decade.
But this release brings enhancements to PerfNav’s tactical features, too. It’s worth noting that PerfNav is not just a strategic tool that’s used once every few years for capacity planning and modeling purposes. While it has a reputation as a strategic tool, it can also be used to track down daily performance issues that crop up during the course of business.
In version 17, MPG bolstered the “What’s Different” functionality, which is a handy diagnostic tool for identifying the cause of performance problems. New features include a new Performance Report Card that lets the user compare two dates; a new infrastructure analysis capability to identify what changed in the hardware or the operating system; and a job analysis capability to identify which jobs are consuming resources.
These new capabilities can help operations people track down the causes of problems, such as that problematic PTF that you told the nightshift operator to apply. “Without a tool like PerfNav and the before and after function, you do a lot of hunt and pecking to figure out what the hell happened,” Young says. “With the software, you can enter a date and look at a graph of any performance metrics–disk arm utilization, CPU utilization, etc.–and see what was going on before and what was going on after. And by figuring out what physically happened in the environment, you can say ‘Ah, that’s a causal relationship.'”
MPG also shipped version 7 of Power Navigator, a similar tool designed for tracking performance in AIX- and Linux-based Power Systems environments, as well as Solaris and HP-UX environments. The Boulder, Colorado, company is seeing more demand for Power Navigator functionality among IBM i customers thanks to the higher reliance on the VIOS virtualization layer, which runs on AIX.
“Almost all the changes and enhancements we’ve made to the product over a pretty long period have been largely at the behest of customers. We haven’t shifted our focus at all,” Young says. “We talk about and demonstrate the capacity planning and sexy stuff at tradeshows. But we’ve added over the years a tremendous number of features to aid system administrative people and support the diagnostics functions, because if you go from nothing to trace in an AS/400 world, you’re taking a pretty big leap off a cliff.”
PerfNav version 17 is available now. For more info and downloads, see www.mpginc.com.