VMware Essentials Info?
-
@NerdyDad said in VMware Essentials Info?:
@Shuey Then you should be good on that front. Can you cluster all 3 of them together with HA? That way, incase one host fails, then you have 2 more to fall to.
Clustering is highly overrated. Almost no SMB needs it, and it adds a ton of problems with MS licensing.
-
Good deal - sounds like the original three-host idea is the way to go then (no Essentials Plus needed
). -
@Shuey said in VMware Essentials Info?:
We currently have a paid VMware Essentials license and we're only using two physical ESXi hosts. As far as I can tell, I should be able to add one more ESXi host in our environment and also add it to our vCenter server.
Reason I ask about this is because I'm considering this train of thought: If one of our existing two hosts ever dies, I don't wanna be scrambling to get a new server up and running, and I don't wanna be screwed on resources since one of our existing hosts wouldn't have enough to host all of the guest VMs from the other existing host if it took a dump. I've thought about building out this third host with at least enough resources (physical CPUs, RAM and local storage) so I could safely restore guest VM backups from either of our existing hosts to the third host. And in the meantime, the third host could be used as a host to do testing on (so it's not just sitting idle while waiting for judgment day to arrive, lol).
Thoughts?
So the problem here is that your current gear cannot handle the workload. You should invest minimal cash into upgrading that instead of purchasing an entire server.
Purchasing an entire server here seems a waste of cash to me.
-
What is the constraint in restoring everything onto a single host if one dies? Would you just be out of storage? That would be my guess since you could likely tune down RAM and CPU to crawl along until one host can be brought back.
Remember you only need enough for one host to be functional if you have a host failure. But it depends on what performance expectations are from management in that scenario. I'd at least ask the question.
-
@JaredBusch @NetworkNerd The RAM slots are maxed out on each of these servers (192GB each, and they can't be upgraded any further), and as NN eluded to, the storage is where we would not have enough resources. To make things worse, the storage on these two existing servers is an MSA60 (which are super old, way out of support and prone to issues). The third server I'm wanting to add is something that we already own; it just needs more RAM and more local storage added.
-
How old are these hosts?
-
Sounds like we have an IPOD situation here with eminent demise. What would happen if the data on that SAN was lost?
-
@NerdyDad That's one of the reasons I want to build out this nice brand new DL380 Gen9 (local storage which will be more dependable than the old crappy MSAs, and I can put 256GB or more of RAM in it). And if either of the existing hosts takes a dump, I have backups of every guest VM that I could easily restore if need be. The existing hosts are a DL360 G7 and a DL360 G8.
-
Have you considered a hyperconvergence solution, such as @scale?
-
@NerdyDad I have, but that takes me back to the situation with upper management: Anything I consider implementing will involve asking them to shell out money they don't wanna shell out :-S. Another thing I've considered is building out two "new" VMware hosts (one bigger/better hardware) and leveraging the free version of StarWind VSAN, then migrating all of our data to the new servers. That of course would be a huge undertaking since I've never done anything like that, and I've never used or read up on SW VSAN.
-
@Shuey with your issue of having an IPOD, then yeah, spin up the new server, migrate all the VM's over to it during maintenance windows (no need for live migration), then shoot the old hardware.
-
The MSA60 is not a SAN. It's DAS. I don't see IPOD as what the environment has at all. The way I read it was each server has a MSA60 attached.
-
@Shuey said in VMware Essentials Info?:
@NerdyDad That's one of the reasons I want to build out this nice brand new DL380 Gen9 (local storage which will be more dependable than the old crappy MSAs, and I can put 256GB or more of RAM in it). And if either of the existing hosts takes a dump, I have backups of every guest VM that I could easily restore if need be. The existing hosts are a DL360 G7 and a DL360 G8.
You mentioned the MSAs are out of support now. What about the servers themselves? My assumption is yes, but I wanted to ask.
Is this just a single-location business, or do you have gear at other locations?
-
@NetworkNerd Yep, the servers are also out of support. The hosts are located in separate data centers (different physical sites across town, in a network that's all connected by fiber).
-
I think it's really time to try and quantify the business impact of losing a host or the storage array attached to the host. What would it cost the company if a host tanked? How much of what was on that host could you restore in a reasonable amount of time (whatever your RTO is), and how much of it would not be able to go anywhere because of lack of space?
Could you at least jot down the VMs on each host and their respective roles to show your boss? Then explain what happens when a host dies. Then drive the point home that nothing we have is under warranty, which means we're looking at eBay or scrambling to band aid something to get the host / its array working again. That will likely delay recovery time. There is value in running vendor supported hardware.
-
If you're really concerned about budget, you could start researching gear pricing on your own from somewhere like XByte at a fraction of the cost of new gear. And they will warranty it for you for years to come. Just make sure it is on VMware's HCL, of course. You could price out new gear as well as the refurb gear from Xbyte and then throw in the refurb gear if something new is shot down due to cost.
But, none of that matters if you don't have some idea of how to quantify that downtime in terms of business impact and dollars. The powers that be may not see the value without some rough numbers.
-
@NetworkNerd Great info to put things in perspective, thank you. A couple things to note here too:
- I'm not the manager of the IT department, so I don't have the ability to talk to upper management about this stuff. I just do my best to convey the info and impact to my "IT Director" and let him know the dangers (in writing), and then try to let go of it so I don't stress myself out.
- Upper management is SO cheap that there have been multiple times where I've shown them "deals" that we could purchase from various vendors, but management says that I HAVE to get them quotes from Amazon (because they "get points for shopping on there, and then use the points for 'free purchases'"). It's hard to work with bean counters who don't even know how to logically count beans... :-S
-
@Shuey said in VMware Essentials Info?:
Upper management is SO cheap that there have been multiple times where I've shown them "deals" that we could purchase from various vendors, but management says that I HAVE to get them quotes from Amazon (because they "get points for shopping on there, and then use the points for 'free purchases'"). It's hard to work with bean counters who don't even know how to logically count beans... :-S
Time to start working on that resume.
-
@Shuey said in VMware Essentials Info?:
@NetworkNerd Great info to put things in perspective, thank you. A couple things to note here too:
- I'm not the manager of the IT department, so I don't have the ability to talk to upper management about this stuff. I just do my best to convey the info and impact to my "IT Director" and let him know the dangers (in writing), and then try to let go of it so I don't stress myself out.
- Upper management is SO cheap that there have been multiple times where I've shown them "deals" that we could purchase from various vendors, but management says that I HAVE to get them quotes from Amazon (because they "get points for shopping on there, and then use the points for 'free purchases'"). It's hard to work with bean counters who don't even know how to logically count beans... :-S
What kills me in all of this is why, if they are so cheap, are they paying for equipment to be in a datacenter? It's like they are currently paying for climate controlled storage for some junky furniture with rips that is stuck together by duct tape.
-
@NetworkNerd I feel your pain, believe me. A couple of examples of how misguided they are (and stubborn - I add that because I've expressed the dangers of various decisions they've made, and my input has fallen on deaf ears multiple times....):
- There are two main data centers. One of the main data centers is stored in a "janitor's closet/storage room" and there's no lock on the door O_o... The building that this data center is located in has no alarm system! The other main data center is cooled by a wall-mounted AC unit. And prior to the AC unit, the room was only cooled by a standard AC vent from the main AC unit that runs throughout the entire building. This data center used to overheat all the time, and the fix that the previous IT Director had for this issue was to continually (pretty much 3-4 days per week) "open the main doors and setup some floor fans to blow the air out of the room" O_o...
- Another example: When I first started working at this place, they didn't have ANY spare PCs built and ready to go in case of an emergency. It was like pulling teeth to convince the previous IT Director to allow me to build at least one spare for each site, but I finally talked him into it.
