The Textbook Things Gone Wrong in IT Thread
-
@DustinB3403 said:
The reason I ask is so that should something afflict the VM C partition that I have some way to recover more rapidly that our Buffalo drive.
You might want a faster restore mechanism. Is your file server currently a full 4TB? How do you recover currently?
-
There are 2 servers acting as file shares. The backup mechanism is via ShadowProtect
If I were going to propose this I would scale up the CPU and RAM to the max that the board can support as I'd also say virtualize everything onto this host. to consolidate our server footprint.
-
Well maybe not the maximum.
-
@DustinB3403 said:
There are 2 servers acting as file shares. The backup mechanism is via ShadowProtect
They are very good for file backups!
-
The data partition could be backup via ShadowProtect.
That or I scale up the CIFS server that is being used on our small XenServer to backup the few less critical VM's I have running there to be large enough to hold 12 TB of data.
I'd probably have to build one for that purpose as well as trunk a few NIC's to get a good throughput.
-
The goal is to get off of equipment that is at its EoWarranty.
-
And these servers have 1-2 external drives attached as backup to them already. There isn't much internal storage on these machines.
-
@DustinB3403 said:
The goal is to get off of equipment that is at its EoWarranty.
While virtualizing everything server side.
-
Scaled up the the 10 Bay Chassis, 2 X Intel Xeon E5-2660 2.2GHz/20M/1600MHz 8-Core 95W, 128GB (8x16G) DDR3 ECC RDIMM and eight Samsung 850 EVO 2 TB SSD's total price is $10742.92.
Which is probably still cheaper than what the MSP will offer as a solution.
-
That is where one, single, huge, new, well warrantied, well supported enterprise box with local storage would shine. Consolidate, save money, put the investment all in one place and replace everything old and crufty all at once. Doesn't just fix, like everything, now but it eliminated technical debt and puts you on a solid road for the future continuing to save money, have more and do less.
-
@DustinB3403 said:
Scaled up the the 10 Bay Chassis, 2 X Intel Xeon E5-2660 2.2GHz/20M/1600MHz 8-Core 95W, 128GB (8x16G) DDR3 ECC RDIMM and eight Samsung 850 EVO 2 TB SSD's total price is $10742.92.
Which is probably still cheaper than what the MSP will offer as a solution.
Especially if you are thinking of it as the COMPUTE node AND the storage node all in one.
-
That's the only way it would work in this environment. We don't have any hosted applications such as an ERP, or Virtual PBX. So it would all get moved onto a larger unit.
Which consolidates the entire company to a single VM Host which has room to grow if need be.
The trouble IMO is the price on those SSD's they seem so expensive when compared to the cost of the Host.
-
And down the road (6 months maybe) purchase a second matching unit for HA which we really want.
-
@scottalanmiller What does he do in the case of an earthquaek (or a fire?) ... It takes out the whole compute & storage infrastructure... but what about his backups?
Edit: Ignore this comment, Dustin said what I was thinking.
-
@DustinB3403 If you use 2 x Hyper-V nodes + StarWind, or use XenServer, you can get the HA bits (shared / synchronized storage) thrown in for free without having to pay the VMware tax.
-
@DustinB3403 said:
Which consolidates the entire company to a single VM Host which has room to grow if need be.
That's the standard model for SMBs. There should only be one host unless there is so much disaster recovery need based around hardware failure that a second host is warranted in which case you simply have two.
-
@DustinB3403 said:
The trouble IMO is the price on those SSD's they seem so expensive when compared to the cost of the Host.
We only showed you SSDs as an example of how much overkill we could do for cheaper than the solutions you were currently being saddled with. No reason to assume that you need SSDs.
-
@DustinB3403 said:
And down the road (6 months maybe) purchase a second matching unit for HA which we really want.
If HA is warranted, yes. But your current environment and everything being currently considered are lower than SA (standard availability) and are actually LA (low availability.) So you know, without the slightest doubt, that HA is not needed in any way whatsoever.
-
@dafyre said:
@scottalanmiller What does he do in the case of an earthquaek (or a fire?) ... It takes out the whole compute & storage infrastructure... but what about his backups?
Edit: Ignore this comment, Dustin said what I was thinking.
It addresses that better than anything else that they are considering. So this isn't a problem at all.
-
The biggest issue here is we have people working globally, 24/7 all coming back to the main office. If this server failed for any reason all functional service would stop until it was back up and running.
Granted I think a second unit is reasonable for the cost of $10G its my boss who has to sell it.
