Getting a crash course

NASHUA, N.H. - Users plagued by Windows NT and 2000 server or workstation crashes may find help in software Alexander LAN will roll out this week that spells out the cause of a crash and how to fix it.

Alexander System Protection Kit (SPK) for Windows automatically diagnoses and analyzes Windows server and workstation crashes, the firm says. The software, which runs on a Windows NT/2000 servers or XP workstations, isolates the cause, records it and alerts the network manager via SNMP traps or e-mail.

When a server crashes, the software captures data about the state of the system at that moment. After the server or workstation restarts, the Alexander software reactivates, processes the system's memory dump and alerts the manager as to the likely cause of the problem.

What's the cause?

Determining the cause of crashes has always been the purview of savvy managers and developers who understand and interpret system debug programs or memory dumps. But most network managers have neither the time nor the skills.

"For those few network managers who do [understand debugs or memory dumps], they really don't have the time," says Steven Snyder, CTO at ISG, a systems integration company in Stafford, Va. "When a machine has a problem, you don't have the option of taking time to figure out what went wrong and try to fix it. You have to get [the server] back online and get people working again."

SPK for Windows includes a Windows-based management console; the SPK Center component; and individual server and workstation agents. The agents, residing on a server or workstation, report a crash to SDK Center, which analyzes the crash and makes recommendations for fixing it. IT managers view SDK Center through the management console.

Other vendors write crash-dump analysis packages that let IT managers manually reconstruct what could have caused a crash. Crash dumps consist of information about the state of the machine at the time of the abnormal end, or termination, of tasks.

Sun provides a similar program, OS Crash Dump Analyzer Module, for Solaris servers that retroactively checks the status of the operating system, prints a report that isolates the cause and alerts the administrator.

A Novell debug mode lets system managers determine the cause of crashes.

Only Alexander SPK for NetWare proactively decides when a Win 2000, NT or NetWare server is going to crash and corrects it, if possible.

SPK for Windows is expected to be available next month for $500 for one server and five workstation licenses.