Oct. 2, 2006

While IT personnel in the B2B segment can customize some data which is possible to populate on any given IT dashboard, efficient data center diagnostical procedures begin with:

1) Severity By Server: This metrics broke down IT problems and performance at an individual server level. You can drill down into each server, see just how it is performing, and what the problems associated with it are. If your IT environment is heavily reliant on server farms and other clusters that don't always make it easy to find the weak link in a chain, this is the kind of visibility you need.

2) Popular Ops Flows: Here you see what components of your technology stack are giving you the most trouble over a rolling 7-day time window. For example, in the demo, there were 376 infrastructure alerts and 325 application alerts. Color codes let you know how many of these alerts have been A)automatically fixed by your IT tools; B)diagnosed but not fixed by your IT tools; and C)neither fixed nor diagnosed. The system also lets you know the average run time on these alerts.

3) Outcome By Alert: Here the system breaks out the result of problems. Possible results are: slow page load, service outages, broken links, missing texts and images, and 500 errors. Once again, you can see how many of these problems were fixed, diagnosed, or unaddressed by your IT tools.

4) Type By Severity: Here you can see the severity class (warning; error; informational; critical; and fatal) by server type (e.g., Windows, Solaris, Linux).

Understanding what's going on in a B2B data center is a first step to doing something about suboptimal processes, and iConclude also has tools to help you create diagnostic and repair processes.

Smith showed me the interface, which looks like Visio and allows you to drag-and-drop the components of your approach.

For example, you can drag boxes marked "send health report," "device diagnostic," and "filesystem diagnostic" into various relationships with each other without having to write a line of code.

Obviously, this is something that a technical employee will be doing, but even so it takes up less of their time thanks to the no-code, drag-and-drop approach.

Most of the time, though, you won't be tweaking diagnostic and repair flows manually, but running automated processes through iConclude.

The product keeps a running log in which it will go through smaller steps (e.g., check connection thresholds, iterative CPU diagnostic, get processor count/status, total memory, general diagnostic, free memory, total virtual memory, &c.), used a color-coded system to let you know which of the components is healthy and which failed.

It also uses text messaging to offer you further pointers.

For example, when the "evaluate expression" component of an automated run failed, iConclude recommended a procedure for solving the problem.

By the way, if you think that data center problems don't leave the data center, you're mistaken. Smith pointed out that, for customers like Bristol Myers Squibb, the iConclude product has Sarbanes-Oxley (SOX) implications, as the audit trail it creates is proof that the company has instituted certain compliance processes.

The iConclude demonstration showed us that, thanks to the visual power of the interface on the front end and the increasing back-end sophistication of IT automation tools, you can get good visibility into your data center and automated a lot of the diagnostics and problems that arise in that environment.

Source: Line 56