I've got a specification for shared storage which is claimed to be suitable for vmware clustering
Axus Yotta Mini Sports FC-SATA-II YI-06SApEF4P-D2 - 1
FC HBA Qlogic QLA2440-CK - 2
SFP 4Gb - 2
Cable opt. LC-LC 1m, Duplex, 50/125micrometer - 2
Am I in the right way?
Thanx in advance
Did you check the hardware compatibility list?
If this is just a play server or something you are using for testing, go for it. If you plan to use it for something you want to keep... I wouldn't do it.
Thank you for fast reply. May you be more detailed? I have checked the vmware-io-guide. The HBA model is there, but not the disk array, of course.
I've heard from my local server builder purists that there is no true HA in this solution due to only one FC controller in the storage box - and I answered them that Vmware HA cluster is some (defined by the vendor) software configuration running and functioning and not more.
My demand is to rerun automatically my critical VMs that have became unavailable on their host. Period here. I have time to restore systems from backup, but due to highly experimental virtual environment my critical VMs tend to fall, and I want to give'em a second chance.
"Thank you for fast reply. May you be more detailed? I have checked the vmware-io-guide. The HBA model is there, but not the disk array, of course. "
Well SATA isn't supported, but many say it works fine. So if you are ok with it, then do it. If your HBA is there, you at least have some support.
"I've heard from my local server builder purists that there is no true HA in this solution due to only one FC controller in the storage box - and I answered them that Vmware HA cluster is some (defined by the vendor) software configuration running and functioning and not more."
Bah, tell them to read the ESX Advanced Technical Design Guide. It explains how HA works, or point them to the knowledge base on VM ware.
1 FC doesn't matter. HA comes from a VC control center (which is not on the ESX server itself) and it manages the cluster and HA by virtue of the NIC or Service Console. So technically if you have multiple NICs and failover for that, you do have HA.
"My demand is to rerun automatically my critical VMs that have became unavailable on their host. Period, no more. I have time to restore systems from backup, but due to highly experimental virtual environment my critical VMs tend to fall, and I want to give'em a second chance. "
This will give you what you want. Although I would still get SAS storage not SATA, because if this is to be the second box, you don't want it any different then your primary. The idea is to be identical, in cases where there is a problem. This box could be used in lieu of the other box, and how knows maybe you will expand one day, you want this to be an ideal box, not "just in case"
Well SATA isn't supported, but many say it works
fine. So if you are ok with it, then do it. If your
HBA is there, you at least have some support.
It's really untrue, and disinformative, to say "SATA isn't supported". No, that's a very untrue statement indeed. Sure, there's plenty of SATA \*CONTROLLERS* that may not be supported, but if an array that is supported is capable of using SATA disks, that is a 100% fully supported configuration. Heck, take a look at something like EqualLogic, a solution that is fully supported on the SAN HCL... They only use SATA in some of their units; notably the PS300E, which is one of the more popular models.
The correct thing to say here is that you should check any storage arrays against the VI3 Storage / SAN Compatibility Guide, located here: http://www.vmware.com/pdf/vi3_san_guide.pdf
"I've heard from my local server builder purists that
there is no true HA in this solution due to only one
FC controller in the storage box - and I answered
them that Vmware HA cluster is some (defined by the
vendor) software configuration running and
functioning and not more."
Bah, tell them to read the ESX Advanced Technical
Design Guide. It explains how HA works, or point
them to the knowledge base on VM ware.
1 FC doesn't matter. HA comes from a VC control
center (which is not on the ESX server itself) and it
manages the cluster and HA by virtue of the NIC or
Service Console. So technically if you have multiple
NICs and failover for that, you do have HA.
When the new ESX ATDG comes out, perhaps.. But, that's a way's off. Let's take a step back and look at what was said, however. While HA is going to help with availability of critical (and no so critical) virtual machines, it's a long way away from being the perfect solution. If your choice in storage only has a single controller in it, and that controller fails... Not HA or anything else is going to bring those VM's online any faster than replacing that controller. Keep in mind that your VI is only going to be as stable and robust as the hardware infrastructure you put in undernearth it...
"My demand is to rerun automatically my critical VMs
that have became unavailable on their host. Period,
no more. I have time to restore systems from backup,
but due to highly experimental virtual environment my
critical VMs tend to fall, and I want to give'em a
second chance. "
This will give you what you want. Although I would
still get SAS storage not SATA, because if this is to
be the second box, you don't want it any different
then your primary. The idea is to be identical, in
cases where there is a problem. This box could be
used in lieu of the other box, and how knows maybe
you will expand one day, you want this to be an ideal
box, not "just in case"
Again, it all comes down to what you're putting in place. EqualLogic makes a very robust and performance-minded array based on SATA drives. Many folks on these here forums attest to that (note: I don't have any EQL arrays, I don't work for them, etc.. I just have heard they have a great rep, so I'm using them as an example).
I think it should be noted that while it's better to have matching hardware to take advantage of things such as vMotion and DRS, HA does \*NOT* have a matching hardware requirement on it, so you can have two disparate hardware types in place if it's just a precaution for potential hardware failures.
SUPPORTED doesn't mean it won't work, VM Ware will not SUPPORT it technically that's ALL I was trying to say. I could use IDE hardware, on a plain desktop machine and get ESX to work, that's not supported, but it doesn't mean it WONT WORK.
Nobody said this wasn't a viable solution. And he was asking advice, and if you look at my response I did ask if he looked that the HCL.
You should be a politician, you would be a great grand stander . . . .
1 FC doesn't matter. HA comes from a VC control
center (which is not on the ESX server itself) and it
manages the cluster and HA by virtue of the NIC or
Service Console. So technically if you have multiple
NICs and failover for that, you do have HA.
<<When the new ESX ATDG comes out, perhaps.. But, that's a way's off. Let's take a step back and look at what was said, however. While HA is going to help with availability of critical (and no so critical) virtual machines, it's a long way away from being the perfect solution. If your choice in storage only has a single controller in it, and that controller fails... Not HA or anything else is going to bring those VM's online any faster than replacing that controller. Keep in mind that your VI is only going to be as stable and robust as the hardware infrastructure you put in undernearth it...>>
What are you talking about? I do this \*NOW* not way off. Are you just saying stuff, or actually responding to the text?
Are you responding to me or him? Direct this to him, I don't need this solution, thanks.