How We Tested vSphere 5

We did a fresh install and mostly upgrades to existing VMware vSphere 4.1 (patched) resources in our lab, which contains several Dell Servers and a DLink GB switch. It's connected in turn to our network operations center at nFrame in Carmel, Ind., 78 miles away via Comcast Business Broadband. 

In the NOC are several servers including three primary test servers, an HP 585 (four AMD CPUs with four cores each), HP 580 (four Intel CPUs with four cores each), and Dell 1950 (two Intel CPUs, four cores each). The servers are connected via two virtual LANs on an Extreme 10GB switch. Also connected is a Dell Compellent 15TB SAN. In turn, we used a Dell 1950 (eight cores) to host and control our infrastructure.

We tested upgrades, storage features, and fault detection (manual drive removals, constant restarts, and other induced failures). We also used traffic generators and CPU "attack" programs to simulate graduated loads) to test the triggers needed for fault detection, VM load management through target resource availability, and fault tolerance (runaway CPU detection).

To continue reading, register here to 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 did a fresh install and mostly upgrades to existing VMware vSphere 4.1 (patched) resources in our lab, which contains several Dell Servers and a DLink GB switch. It's connected in turn to our network operations center at nFrame in Carmel, Ind., 78 miles away via Comcast Business Broadband. 

In the NOC are several servers including three primary test servers, an HP 585 (four AMD CPUs with four cores each), HP 580 (four Intel CPUs with four cores each), and Dell 1950 (two Intel CPUs, four cores each). The servers are connected via two virtual LANs on an Extreme 10GB switch. Also connected is a Dell Compellent 15TB SAN. In turn, we used a Dell 1950 (eight cores) to host and control our infrastructure.

We tested upgrades, storage features, and fault detection (manual drive removals, constant restarts, and other induced failures). We also used traffic generators and CPU "attack" programs to simulate graduated loads) to test the triggers needed for fault detection, VM load management through target resource availability, and fault tolerance (runaway CPU detection).

We also tested vSphere Storage DRS using the Compellent SAN, although the pieces needed to do a full object migration were unavailable to us at test time.

Return to main test.

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