Hey iDoctor, Can i Get a Walker?

November 19, 2013 Alex Woodie

IBM is currently beta testing a new piece of software called the HMC Walker that’s designed to help Power Systems admins pinpoint LPARs that aren’t running at optimal speed. When an admin finds an underperforming LPAR, the HMC Walker interface makes it easy for the admin to drill down into the iDoctor tool to get more details on the problem.

The new HMC Walker offering gives administrators a high-level view of all their physical systems and the LPARs that are attached to them via the hardware management console (HMC). The software works with LPARs of all types, including IBM i, AIX, Linux, and VIOS, according to IBM’s iDoctor Web page.

“HMC Walker is designed to capture high-level performance and configuration data via HMC commands and present it back to the user in a way that will be easy to use and understand,” says Ron McCarger, who works in the IBM i Global Support Center in Rochester, Minnesota. “The configuration data can be used to keep track of configuration changes or maintain a history of changes to the HMC configuration over time.”

HMC Walker will enable admins to view LPAR usage information in a graphical manner. For example, if LPAR_45 is consuming more than its share of CPU, it will show up clearly in the HMC Walker user interface, allowing the admin to easily identify it.

“Currently you can drill down from a managed system graph to a graph showing all LPARs for the selected managed system,” McCarger tells IT Jungle via email. “In the future, from a selected LPAR, additional drill downs will be made available that will utilize existing iDoctor components, like Collection Services Investigator or possibly Job Watcher, which is like Collection Services but more detailed.”

In this view of HMC Walker, you can easily see that the pink LPAR is sucking up more than its share of CPU. Bad LPAR!

HMC Walker currently gets performance data using the lslparutil command. “This gives us CPU and memory statistics for all managed systems and LPARs at a high level,” McCarger says. “You can configure collection to happen automatically using scheduled ‘batch’ tasks that will go out and collect the data every day at the desired time down to your ‘analysis database.'” Currently, that analysis database is DB2 running on IBM i, or Oracle Express running on a Windows PC. The company plans to support the database on AIX in the future.

IBM is developing HMC Walker to make it easier to view performance data for all managed LPARs in a Power Systems environment, which is an item on the Large User Group’s wish list. It also has the capability to expand the types of performance metrics available to IBM i customers, McCarger says.

If the HMC Walker needs a hand, maybe it can call in this guy to pinpoint LPARs.

“On IBM i, that gives us the capability to view wait buckets, CPU utilization, I/Os, JVMs, etc over time or rank the jobs within by the desired metric,” he says. “The drill downs will vary depending on the type of LPAR selected and whether or not data is even available. NMON data will be used as the drill down of choice for AIX/Linux systems and VIOS Investigator’s NMON collection mechanism will be used for VIOS. It could also give the user the option to view a VIOS Advisor report for the desired LPAR or add options to collect that on a regular basis [once per day for example].”

The open beta for HMC Walker started in early October, and will run into the new year. HMC Walker should become available before the third quarter of 2014. For more information or to request to become part of the beta, see www-912.ibm.com/i_dir/idoctor.nsf/hmc.html. You can also view a YouTube presentation by McCarger.