Profiling your infrastructure

Last week'scolumn explored how Onaro SANScreen Foundation can bring more control to your datacenter by creating a comprehensive view of servers, storage, and applications. That single overview helps monitor vital signs and enforce SLAs.

[ MarioApicella's column is now a blog! Get the latest storage news from the Storage Adviser blog. ]

This week, I'm turning the spotlight on another management application: Akorri and its BalancePoint Suite 1.3. Akorri promises end-to-end performance monitoring and analysis with the additional twist of a built-in analytics engine that helps drafting an accurate profile of your datacenter resources.

Akorri began pitching its applications just last week, but the company already has a group of customers -- some of which have been using the suite since October.

You can read more about Akorri's raison d'etre on founder Rich Corley's blog. He's written some interesting flashbacks on Pirus Networks, a storage networking company that Corley founded which Sun Microsystems later acquired.

Akorri delivers the BalancePoint Suite preinstalled on a 2U appliance running Suse Linux. In a typical deployment, you connect the appliance to your Ethernet network and let the data collection begin.

Thanks to a variety of standard interfaces such as WMI, SSH, ODBC, and SMI-S, BalancePoint will collect performance data about just any server storage device and application in your network. Akorri says customers should get some significant information on their systems' status after only a few hours or, depending on the business, after running the discovery process long enough to include activity during peak hours or a weekend.

BalancePoint discovers most of the objects (servers, application, and storage devices) and their connections automatically, but offers the possibility of adding objects manually if necessary. What has to be entered manually are the performance requirements of each application, data that BalancePoint compares with the data it collects to identify bottlenecks and problem areas.

All that information lives in an embedded database, which BalancePoint's analytics engine examines to create application profiles inclusive of critical performance elements, such as usage of CPU, memory, and storage.

The built-in analytics are probably the most interesting aspect of BalancePoint. These tools identify problem areas that are otherwise not immediately apparent, such as a contention on shared resources such as storage array or a network path.

The first screen I saw during a live demo of Balance Point was a dashboard presenting an overview of the datacenter status and listing applications with a history of poor performance. A chart on the same window pinpoints the general area to investigate, absolving storage arrays and singling out the applications and servers as possible culprits.

Clicking on an application brings up more performance-lag details. For example, choosing Exchange revealed a memory constraint on its machine, a problem that's usually much easier and less expensive to fix than, say, adding another storage array and spreading the mail volumes across more spindles.

In addition to quickly identifying a performance weakness from historical data, BalancePoint can predict the impact of changes and suggest where adding new resources (say, more storage) would be most effective.

BalancePoint doesn't come cheap, however; Akorri suggests a price between $250,000 and $350,000 for deployment to a large datacenter. Nevertheless, for companies facing serious bottlenecks and resource management issues, its potential benefits could be too good to pass up.

Join me on The Storage Network with questions or comments.