Optimized virtual interrupt delivery in ESX 4.0 improves the performance of some Linux guests. On an I/O completion, previous versions of ESX scheduled the vCPU that originated the I/O request. If this vCPU was found to not be the target vCPU chosen by the guest for further processing the I/O completion interrupt request, the VMM forwarded the I/O completion to the correct target vCPU with an additional inter-processor interrupt. ESX 4.0 avoids the need for these additional interrupts by predicting which vCPU is the target of a particular IRQ and delivering the I/O completion to it instead of to the originating vCPU. This reduces the CPU overhead substantially.
- RT @SiliconANGLE: Google is investing $310M in a new Belgian data center with an on-site solar farm ift.tt/2Eylb66 1 month ago
- RT @amyengineer: Cannot stop laughing… :) twitter.com/ThePracticalDe… 2 months ago
- RT @vswitchzero: [ICYMI] An in-depth look at VMXNET3 buffer exhaustion and packet loss. How and when you should tweak the buffers! @vmware… 2 months ago
- Guys if you have laying around a #SnapGear #SG580 aka #McAfee #UTM #SG580 just ping me I’ll buy it very good price! #vExpert 2 months ago
- RT @LukasStefanko: "Our data breach happened using a highly professional attack with sophisticated social engineering" https://t.co/eFl6Xos… 3 months ago
How To Troubleshoot… on Chunk Size Of a RAID0 Volume O… PiroNet on It All Started With This … Gorka on It All Started With This … An administrator not… on Ballooning And Hypervisor Swap… revate on Installing Oracle Database Cli…
Views expressed here are mine, they are not read or approved in advance by any company and don’t reflect the views of my employer, my employer’s business partners, or clients. I am solely responsible for all content produced here. No information provided here was reviewed by or endorsed by my employer or any other vendor or organization. This is my own blog. Comments are moderated!