1

Hello all,

I need a sanity check on a new server purchase. We’re UK based, i’ll pop this up front as sometimes pricing and availability varies massively from country to country and I’m mostly looking for general advice.

We’re looking at buying another HPE server to add to our growing onsite VMs, not looking at cloud based right now or possibly ever for these VMs.

We’re going to be running ESXi 8.0 on the bare metal and a variety of VMs on it. Mostly I suspect they will end up being smaller services running on ubuntu as we chip away at the older monolith stack splitting it up (not my job, yey!).

Currently we’re looking at the HPE DL380 Gen10plus with 2xXeon silver 4310 processors, 128gb ram & 24tb of hard drives (after raid5 20tb) plus an NVMe boot drive. I’ve added a 10gb network card to future proof that angle although we don’t have a 10gb backbone right now. Total plus some extended licencing is around £11k.

Our existing servers are both running similar specs with older CPUs about half the hard drive space and the same ram and are currently running at about 80% capacity on RAM and HD space (although i need to review the VMs as i suspect some of them are over provisioned).

My current thinking and internal debate is should i be looking at 256gb of ram rather than the 128gb to try and future proof further or is this just a folly? Is there any other servers we should be looking at instead? Is there anything else i should be considering? This is the first server i’ve had to purchase and want to make a good decision as i’m going to have to live with it, haha.

6 Spice ups
2

We will most likely be going to the Cloud for our next Infrastructure refresh but currently we are still running on Dell VRTX kit and it has been very reliable for us. I agree it is probably best in your case to go with 10gb network and 256GB Ram rather than 128. As for HDD, we started with a half-full chassis but inevitably ended up adding disk later so if your budget is there, fill it up if you think you may end up needing it.

Having said all that, have you done some forecasting your current and future needs? It does sound like you are spending a lot on server hardware, how long will this system be expected to be around?

3

Hi @briser ​, thanks for your response.

Good advice re the HDD space, thanks.

Yea we’ve done some rough forecasting (back of the napkin maths) we’re 90% confident adding in this 3rd server will allow us to load balance as well as being able to reboot hosts as required for patching and guarantee future expansion if required. This new server will likely be in place for 4-6 years. We’re planning on one other new server next year with similar spec to this one but that will be a replacement for one of the existing hosts that is around 5 years old.

1 Spice up
4

Ram you can purchase additional modules whenever, so I wouldn’t be terribly bothered, as long as you have plenty for your current plans. Storage wise, RAID5 is dicing with death on an array that big, the URE value of drives makes it fairly likely you’ll fail a rebuild. Also, if you’re doing VM infrastructures it’d make more sense to use a SAN. For reference our VMWare infra in the office has VM servers with small storage (i forget what because we don’t care) in raid 1 for esxi, then we have 2 SANs that replicate with each other, we can then move VMs from server to server without having to wait for storage migration. If a VM box goes down, vcenter spins up the VMs on another front end box, if a SAN goes down, we fail over to the other SAN. If we’re doing maint, we can just pop a host into maintenance mode and VMWare takes care of moving the VMs for us, we only lose a few pings and can then get on with whatever we need to do, bring the box back, take it out of maint and VMWare uses the load balancing rules to move the VMs back, again, we only drop a few pings.

We do local storage where we’ve got small clients who benefit from running VMs simply to reduce their server footprint and power bills, but if you’re doing multiple servers then might as well take advantage of all the things virtualisation is incredibly cool for.

5

Be careful with that RAM statement - depending on what config / how many slots you take up and what are future needs. I’ve had to replace the full load with higher capacity DIMMs to maintain compatibility.

We run mostly Windows VMs on ESXi, so our hosts are typically 32vcpu / 512gb RAM

1 Spice up
6

Hey OP!

Hats off for doing your research, server purchases are definitely an investment worth considering your options on! If you do end up going ProLiant, I will say there is a ton of flexibility in configuration/scalability available as well as some great support from the HPE team (I’ve also tagged in our HPE experts here in the Spiceworks community in case you wanted to reach out to them directly or if they have any additional advice to share from their end)

Since there are so many options for configuration , I didn’t want to bombard you with any tech-dense resources that may not be helpful but please feel free to reach out if there’s any specific info or resources I can dig up for you to help make your decision, always happy to help! In the meantime, best of luck and congrats on your first server purchase!

@HPE

@calvin-hpe @mark-hpe @lifecoffeegaming