|
Are Two Virtualized Halves of a Nehalem Box as Good as the Whole Thing? Depends on the OS, Apparently
Curious about the impact of virtualization on benchmarking results? Jan Brosowski blogged in SAPonSun about testing two 'half-box' virtual machines on the same box. In one instance, the benchmark used the Sun Fire X4270 running the Solaris 10 OS and Solaris Containers for virtualization, employing half the COPs available in the box (8vCPUs), Oracle 10g, EHP4: 2800 SD-User. In another, the benchmark used the Fujitsu Primergy RX 3000, SuSE Linux Enterprise Server4 10, VMWare ESX Server 4.0 and, once again, 8vCPUs (half the available CPUs in the box), MaxDB 7.8, EHP4: 2056 SD-User. The obvious question, Brosowski writes, is whether two "half-box" virtual machines on the same box perform better than the full box? No, he concludes, adding that the answer is obvious. Because of Intel's Hyper-Threading technology, it's complicated to extrapolate from half of a system to a fully utilized system, he points out. Both benchmarks used 8 virtual CPUs representing 8 of the totally available 16 threads, and not a single CPU. So they will have used 8 cores, but only the first thread of each. Will the use of VMWare mean a significant loss of performance, he asks. Yes, and again, obviously so. He attributes some of the lower performance to the OS, given the clear, demonstrated advantage of Solaris as an OS for SAP. But, Brosowski stresses, the difference is a lower performance of 36%, more than one third. Clearly, a major impact of the virtualization. More Information
Sun-Oracle Deliver World Record on Two-tier SAP S&D
[...read more...]
From the latest issue:
|
|
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||
Print Friendly
Email this article
Virtualization on the half-shell: What happens to performance?
Are Two Virtualized Halves of a Nehalem Box as Good as the Whole Thing?
Copyright © 1998-2009 System News, Inc.
