How we tested VMware's vSphere

We tested VSphere 4.1 using HP DL580 (16 Intel cores) and HP DL585 G5 (16 AMD cores) servers, along with Dell 1950 servers (eight Intel cores, lots of memory) on a switched GBE network, that's, in turn, peered at 100mbps, located at nFrame in Carmel, Indiana.

We tested VSphere 4.1 using HP DL580 (16 Intel cores) and HP DL585 G5 (16 AMD cores) servers, along with Dell 1950 servers (eight Intel cores, lots of memory) on a switched Gigabit Ethernet network, that's, in turn, peered at 100Mbps, located at nFrame in Carmel, Ind.

We tried to upgrade from vSphere 4.0 to 4.1, and found problems that were rapidly remedied by VMware's support team. We tested several of vSphere and vCenter's new features, then embarked on a mission to test the new multiple concurrent vMotion virtual machine transfer capabilities. We found vMotion to be limited, running sequentially rather than concurrently. After much tuning (and aid from VMware tech support), we were able to sustain four VM transfers using vMotion. Had we used a 10GB switch, the full eight concurrent moves may have been possible, we speculate.

By tracking with vCenter, we tested claims that overhead memory use for discrete VMs was more efficient, and were able to verify this but only for 64-bit VMs.

Return to main test.

Insider Shootout: Best security tools for small business
Join the discussion
Be the first to comment on this article. Our Commenting Policies