Jump to content

Looking to go virtual - been suggested the Dell VRTX for everything?


Recommended Posts

Posted

We are trying to move away from our one-physical-server-per-role approach to a virtualised setup, where 2 physical servers running vmware would be sufficient and initially have been looking at a dual HP-Proliant-380 setup with a SAN connected via switches to achieve this which I believe has been a typically standard approach.

 

However, we were made aware of the Dell VRTX server (never really considered dell previously) but this does actually sound too good to be true.. "everything" in a single box including the ability to expand physical servers and have the equivalent to our SAN on it too without compromising on anything?

 

What we don't want to do is seriously consider it if it's got flaws (lesser powered, limited expandability, minimal upgrade potential etc) in comparison to having actual full height, dedicated physical servers. Is there any reason why we wouldn't want to consider the VRTX? Besides the fact that when I look on the spec they still have Sandy Bridge processors only available, is there any "gotchas" we need to be aware of?

 

I was always of the thought "if it sounds too good to be true.." so need to ensure we are aware of any pitfalls if they exist.

Posted (edited)

Its really a mini blade enclosure. The issue is your limited on capacity. If you run out of space for blades and storage you will have an issue. Something really good about these is they are nearly silent and don't throw out a great amount of heat, excellent for branch offices. I don't think it can be rack mounted.

 

I don't think the VRTX has 10GB uplinks. Would be an issue.

Edited by FN-GM
Posted

I had a two node VRTX setup installed back in the summer and I love it!

 

I am only using 10 out of a potential 25 HDD slots in mine and this comfortably holds all of my VMs, file storage, and backup to disk locations.

 

With a small registry change Microsoft HyperV will accept the VRTX Drives for Fail-over clustering which is blisteringly fast when it comes to fail-overs.

 

I've got a single web console for everything now; both of my nodes, the storage, the networking... it is brilliant.

 

Oh, and the VRTX is rack mountable - mine is free standing as my server cab is moving in a year or so but your Dell dealer will be able to arrange a mounting kit.

 

(sorry this sounds a lot like an advert but I have been so impressed with the entire thing and I still have tons of room to expand with another two nodes and shed loads of hard disks)

Posted

If you wanted a larger system for expandability We could do you a HP C7000 Chassis with as many blade servers and storage blades as you wanted (Up to 16 in total) :)

 

That should be an all in one solution for you!

Posted
Looks pretty impressive actually. The one thing that seems to possibly let it down is the lack of 10GbE. If you don't have 10GbE uplinks or plan to in the next few years to though that point is moot.
Posted
We have been looking at these and was very impressed looking at one at BETT. We are seriously considering using one of these to transition our Citrix VDI cluster which is currently spread over three Dell 720's. They really seem to me to be a an impressive product, and I think Dell are underselling them a bit as they seem to be targeting the "branch office remote site no space for a proper server room" market. I'm pretty much committed to getting one for us to install over the summer.
Posted (edited)

According to PC Pro's review you can get 10GbE mezzanine cards.

For networking, each VRTX node has a 10GbE mezzanine card, which links directly via the midplane to an eight-port switch at the rear. This is a basic L2 Gigabit switch and Dell plans to offer a 10GbE version as well.

 

I assume that was the review unit as configured, so may just be an option. EDIT: rereading more closely, your actual networking is limited by that 8 port switch - so it may be that 10GbE is in the near future, then.

 

Our virtual server infrastructure isn't due for replacement for a couple of years yet (been in three) but this is an interesting solution... should make getting the servers & SAN to talk to each other a damn sight easier, anyway.

Edited by sonofsanta
Posted
According to PC Pro's review you can get 10GbE mezzanine cards.

 

 

I assume that was the review unit as configured, so may just be an option. EDIT: rereading more closely, your actual networking is limited by that 8 port switch - so it may be that 10GbE is in the near future, then.

 

Our virtual server infrastructure isn't due for replacement for a couple of years yet (been in three) but this is an interesting solution... should make getting the servers & SAN to talk to each other a damn sight easier, anyway.

 

The coms between servers in the enclosure is 10GB but the connection to an external switch is limited to 1 x 8GB. You could team these up so it would be like a 8GB connection.

  • Thanks 1
Posted (edited)

It's something that on the face of it seems like the perfect solution to us, as we were only looking at having 2 physical servers to virtualise on anyway and this provides provision for 4, plus the higher transfer rates to storage that would otherwise be on an external SAN.

 

I am just finding it hard to get my head around having the entire network.. literally everything.. on a box which doesn't seem much larger than my desktop tower (depth aside).

 

Seems that the max CPU Dell will do is from the Ivy Bridge family, and whether or not the chassis will allow for newer blades to be inserted I am not sure, but everything points to a solution that is just ideal and I am trying to find what is normally the inevitable "but"s yet can't find anything particularly deal-destroying. Could it really be as it seems??

Edited by dgsmith
  • 1 year later...
Posted

We went ahead with it last summer and I have been very pleased with it. We bought one with four servers and a 10GB Ethernet card and migrated all our core servers to it, moving them from XenServer to Hyper-V. We stuffed it with discs and it is also our primary datastore. I replaced a somewhat overspecced and flaky SAN with two windows file servers running on the VRTX and this has simplified things considerably.

 

As a result, we could replace 8U of old Dell servers and 8U of SAN with one 5U box. Massive saving in space, power, cooling and cabling. I think its an excellent piece of kit and would very much recommend it.

  • Thanks 2
Posted
No, we're not that sophisticated. We just have two to spread the load - one is user shares the other is everything else.
  • Thanks 1
Posted
Did you go ahead with this?

 

Yes, like Linfit we did this over summer. Network config on the VRTX has been a bit faffy (getting the vlans done etc) but the kit itself and the savings in cost, space, power, noise and cooling has certainly made it worthwhile. Pretty much everything is in a box smaller than one of our old servers alone!

We are only using 2 blades and half the overall capacity in 2.5" HDDs too so we can in theory expand double what we have now within the same footprint as we currently have.

  • Thanks 1
Posted

I've got one, direct from Dell, 2x M520 nodes and a mix of SAS and SATA 2.5" disks. Hyper-V failover clustering set up. It's going nicely. Switch is a bit of a faff but you get used to it in the end.

 

I got 2x 4port PCIe cards I have mapped to each node, so each node has 8 network cards visible. One of them is reserved for the use of live migration and the other 7 are teamed so effectively each node has 7GB's onto the network. The downside of this is that I have 16 ports on my server room switch dedicated to the VRTX :lol: (It has two management sockets!)

Posted
For those who said switch is a faff (which I had wondered about) - would it be better to get the pass-through if I've capacity on the switches, then all VLANing is just regular switch config?
Posted

Do you mind PM-ing me the number of your Dell account manager?

 

I've raised a query via Dell directly but I've had a callback from someone in India rather than a UK-based account manager.

Posted
For those who said switch is a faff (which I had wondered about) - would it be better to get the pass-through if I've capacity on the switches, then all VLANing is just regular switch config?

 

The only reason I didn't do this is I have VLAN 999 on the VRTX switch for LiveMigration only... On a totally different subnet. Plus you only have 2 NICs per node if you go that route, so with the switch if you have 2 nodes (which seems to be the norm), you loose the use of 4 ports on the passthrough.

 

What's the issue with the Perc card? works in the correct way for me, I've got the cluster storage mounded to C:\ClusterShared or something like that on each node.

Posted (edited)

I've just spoken to my Indian account manager and Indian technical specialist.

 

I need to take off my 'Mr Bigot' hat (although, to be fair, I have had a lot of issues with call centres based in other countries if their English isn't very good). They were both very helpful indeed (though there was a touch of a language barrier).

 

They are going to quote me for the VRTX and also the FX once I've generated some IOPS data using DPAK.

 

I'm also going to speak to Ed at VeryPC to see what solution they can come up with, and also a local Devon-based provider who I've used before. Three options should be sufficient.

Edited by Tall_Paul

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...