How we did it

We deployed EMC VMware ESX Server across three platforms, an IBM x3550, IBM x3650 and IBM HS-21 XM blade (all with twin, four-core Intel CPUs, 16G to 32GB of dynamic RAM with 146GB of SAS disks) and used Windows 2003 Enterprise Server and Red Hat RHEL5 operating systems. We generated 48 virtual servers (16 per physical server).

We connected VMware's SNMP Management Information Base (MIB) to the NetVigil program and commenced using both the MIB and operating-system-specific tests, tracking whether the tests created the expected alarm conditions.

We then tested applications (HTTP, Microsoft Exchange Server, Microsoft SQL Server, Microsoft DNS services, as well as Berkeley Internet Name Domain and Lightweight Directory Access Protocol on Linux) and tracked simulated events with each application, noting NetVigil's ability to correctly track tracking.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

We deployed EMC VMware ESX Server across three platforms, an IBM x3550, IBM x3650 and IBM HS-21 XM blade (all with twin, four-core Intel CPUs, 16G to 32GB of dynamic RAM with 146GB of SAS disks) and used Windows 2003 Enterprise Server and Red Hat RHEL5 operating systems. We generated 48 virtual servers (16 per physical server).

We connected VMware's SNMP Management Information Base (MIB) to the NetVigil program and commenced using both the MIB and operating-system-specific tests, tracking whether the tests created the expected alarm conditions.

We then tested applications (HTTP, Microsoft Exchange Server, Microsoft SQL Server, Microsoft DNS services, as well as Berkeley Internet Name Domain and Lightweight Directory Access Protocol on Linux) and tracked simulated events with each application, noting NetVigil's ability to correctly track tracking.

NetVigil tracked what we did, noting availability and thresholds (memory use and paging) for VMware, and also tracked operating system components (disk and CPU use).

We then used Microsoft's MVS to host several instances of Microsoft Windows Enterprise server and performed the Microsoft-specific tests, noting variance from WMI-produced data. Under both VMware and MVS, we checked simulated business-application-object settings to profile normal conditions for the applications we tested, and proceeded to simulate out-of-threshold circumstances to track alarm notification and correct correlation settings for alarms.

Read more about data center in Network World's Data Center section.