I've read a handful of guides and perused several documents
Yet, I still have some lingering questions about Fault Tolerance (FT). For what its worth I've got a Dell PowerEdge T710 with dual Xeon E5504's (quad cores at 2GHz; no HT) and I'm looking to stand up another server for FT to ensure HA for my CentOS 5 VM.
Question 1a: Does the CPU letter designation (E, L, X etc) matter?
I'm looking to add a system with dual L54XX's (e.g.: L5408, L5420 etc) or a single/dual X34xx (e.g.: X3430) but I want to make sure FT (and vMotion, HA etc) will play nicely in the cluster (with 'Intel Xeon Core 2' EVC mode enabled) consiting of my other server.
Question 1b: What VMware documentation covers this topic?
Question 2a: Does adding systems to the cluster with varying speeds, CPU or core count a cause for concern?
Question 2b: What VMware documentation covers this topic?
The L5420's I'm eyeballing are listed in the guide above clock in at 2.5GHz - my server 2GHz. I ask this based on what I read here: http://itknowledgeexchange.techtarget.com/virtualization-pro/vmware-fault-tolerance-what-it-is-and-h...
Question 3: Are there any supported processors that are not compatible with other supported processors?
Question 3b: What VMware documentation covers this topic?
I've got an E5504 host at home and am looking to stand up another server that's not in the 5500 series (likely 54xx or 56xx) and add it to the cluster. I need to make sure there aren't going to be any issues before investing in new hardware. According to what I saw here, it looks like the 55xx series only plays nce with other 55xx series processors: http://www.gabesvirtualworld.com/list-of-vmware-ft-fault-tolerant-compatible-cpus/.
Many thanks for your time & expertise.
For what is worth, I don't expect to be, and don't want to be, spoonfed. I'm happy to do the research myself if someone can point me to one, or more, documents that will answer these questions.
And *bump*
First, FT and HA are not the same thing. HA will restart your vm after a host-based failure, and FT will store and replay cpu instructions on a second vm that will take over processing if the primary vm fails, so downtime is not occurred.
The letter designation usually indicates power utilization differences, but as long as the CPUs are in the same processor family, you will be fine. The vendors used to put out a vmotion compatibility matrix, and you can search for dell's version of one. Not sure how current it is, but there should be one at least.
The cpu or core count does not matter, only that the processor can use the same EVC profile. The bulk of the load will be for the ft network that will be used to transfer instructions between the primary and standby vm's.
-KjB
Thanks for the responses and for the link. I had seen it before and worked from it in order to get the cluster up.
Thanks also for answering Questions 1 & 2. I'm pleased to hear that physical CPU and/or core-count doesn't matter.
But I'm still in the dark on two issues, although I suspect I already know the answer to one of them
Here's a more definitive list of processor families.
http://www.vmware.com/support/sitesurvey/help_2_5_2.html
Doesn't look like the 5500 and 5400 series combo will work. CPU speeds are typically not an issue, but the tool here will flag anything larger than 400 MHz difference.
-KjB