Alok Kataria , a Systems Programmer & Linux Kernel hacker at VMWare posted to the Linux Kernel Mailing List last month about his experiences benchmarking VMware’s paravirtualization technique (VMI) and hardware MMU technologies (HWMMU) on VMware’s hypervisor.
Their test platform was a Dual Quad Core AMD Opteron 2384 2.7GHz (Shanghai C2), RVI capable running a development build of ESX.
They ran different real world test on a SLES 10 SP2 with a 2.6.16.60-0.37_f594963d-vmipae Kernel and they concluded that in most of their benchmarks the EPT/NPT (hwmmu) setup provides equal or provide better performance compared to VMI.
Below is a short summary of performance results between HWMMU and VMI.
These results are averaged over 9 runs. The memory was sized at 512MB
per VCPU in all experiments.
For the ratio results comparing hwmmu technologies to vmi, higher than 1
means hwmmu is better than vmi.compile workloads – 4-way : 1.02, i.e. about 2% better.
compile workloads – 8-way : 1.14, i,e. 14% better.
oracle swingbench – 4-way (small pages) : 1.34, i.e. 34% better.
oracle swingbench – 4-way (large pages) : 1.03, i.e. 3% better.
specjbb (large pages) : 0.99, i.e. 1% degradation.
In his post Alok tells us that VMware expects that these hardware virtualization features will be generally available in all hardware by 2011. Other reasons for VMWare Paravirtualization’s VMI , such as timekeeping have also left the building therefore VMWare decided that they plan to stop supporting VMI
In short … VMWare has prepared a patch to remove all VMI related code from the Linux Kernel.
The thread on LKML went on noting that current users still might want to have the patches present so immediate removal of the feature wasn’t really the best way to proceed. But some day soon the code will be removed from the standard Linux kernel.
So what does this say about the future of Paravirtualization ?