at what scale does VDI become viable

T

Thuleman

Supreme [H]ardness
Joined
Apr 13, 2004
Messages
5,833
We have had a couple posts about this before about folks complaining that VDI was just too expensive in comparison to just buying individual PCs and be done with it. It's a complex topic to accurately pin down because much of it depends on the actual support cost of individual PCs, but looking at it from a hardware + licensing standpoint, at what scale does VDI become economically viable?

Meaning what (ballpark) capital expense does it take to virtualize 250 or 500 desktops?
 
It doesn't. If you don't take soft costs in to consideration no scale really gets you there right now.
 
That's really unfortunate. I read an article about the power savings VDI provides (~85% on the desktop side), and that's quite significant even at just a couple hundred machines, but that too is "only" ~$20k/year and if it's a quarter million to implement a VDI for a couple hundred machines then it's just in no relation to the benefit.

I think the soft costs are greatly exaggerated in the VMware marketing materials and supporting studies they cite. For example, VMware cites a "recent" IDC study saying that "Organizations deploying VMware View saved on average over $610 per supported end user per year compared with organizations using unmanaged PCs."

Note how they say "unmanaged" rather than "physical"? Which organization of decent size does have unmanaged PCs? Most folks who run Windows on more than a couple dozen machines will run Active Directory to manage clients, already eliminating a large chunk of the potential VDI cost savings.

I don't do desktop support at work but I talked to the guys who do and they said that among 300 PCs they get about 3 a year that screw their machines up badly enough to require a rebuild, the rest of the users doesn't require any help at all. Desktop support costs at our place seem to be very very low, and yes we do run AD.

Just for giggles though, how much are we talking to virtualize 300 desktops? 300k?
 
NetJunkie said:
It doesn't. If you don't take soft costs in to consideration no scale really gets you there right now.
Click to expand...

Yeah, there's other ROI there too with zero clients and financial institutions or anywhere data is sensitive. Taking the cube off the desk home doesn't have any potential risk... taking a desktop with a hard disk in it does.

At some places security is very tight and the workstation software changes a lot. Keeping things patched is critical and not easy to obtain a 100% patch rate. When you start talking thousands of desktops, the soft costs are very real. There might be other ways to justify costs, for instance by elminating the need for a TS farm, etc.
 
For our uses at work, it's about the functionality it provides vs. any ROI.
VDI is not always just about replacing desktops.
 
One of our biggest is user desk shuffles. One moves here, then somebody wants that spot, etc. One user moving = 10 users moving. With VDI, they can move until their heart's content - and we wouldn't have to move a thing.
 
2/3 of that cost is the SAN...I'm really hoping that as SSDs get cheaper, the cost of VDI will drop quite a bit.
 
VDI is only good if you have a huge multinational company and centralise all desktop and server support. That way you can cut down on IT staff in a big way. If you are just looking at cost from an IT perspective it never becomes viable.
 
thefreeaccount said:
2/3 of that cost is the SAN...I'm really hoping that as SSDs get cheaper, the cost of VDI will drop quite a bit.
Click to expand...

Not the case any more. A lot of the cost is in the thin client and licensing. Back-end storage is a lot less these days thanks to things like FAST Cache, linked clones and 3rd party solutions such as Illio. It's not nearly what it was.
 
Jay_oasis said:
VDI is only good if you have a huge multinational company and centralise all desktop and server support. That way you can cut down on IT staff in a big way. If you are just looking at cost from an IT perspective it never becomes viable.
Click to expand...

Not true. We do a lot in health care and other more traditional settings. It's about how you want to manage your environment. I'd rather have people focusing on solving business problems than fighting desktop issues. I was a Network Manager at a 300-person shop and would have implemented a VDI solution if it had been available at the time.
 
Jay_oasis said:
VDI is only good if you have a huge multinational company and centralise all desktop and server support. That way you can cut down on IT staff in a big way. If you are just looking at cost from an IT perspective it never becomes viable.
Click to expand...

False. Again, the benefits of VDI varies upon use case. I take it you're not using it.
 
Brahmzy said:
False. Again, the benefits of VDI varies upon use case. I take it you're not using it.
Click to expand...

thats where you are wrong.

Why not give us some info on when it becomes viable (don't forget the OP is more than likely talking about cost)
 
Well I am indeed talking about cost. We don't deal in secrets or proprietary information such as banks etc, so the security benefit doesn't apply to us. Our soft cost in terms of desktop support are very low. I don't know where market research pulls their numbers from but from what I have been told the majority of our desktops doesn't require any support that would be reduced by VDI over the course of the year.

Our machines are AD managed, and we do have the need to re-image approximately 40 machines twice a year which is done centralized via Norton Ghost server iirc.

The tangible benefits I see are lower power consumption and ...., I don't know, they rest becomes a lot less tangible. Standardized hardware is a benefit though assigning a dollar value to that will be difficult since we have little to no problems with the wide variety of desktops people use. There are some payroll savings in deploying new machines and probably also in the re-imaging of our 40-ish boxes. "Portability" would be a benefit, it will work no matter where in the building people are located.

One main concern is that we are heavily relying on Adobe products from Acrobat to Photoshop, InDesign etc and the performance of those via VDI seems to be lacking at this time.
 
For a typical corporate user, it makes no sense to virtualize the desktop. Tools provided with Active Directory and even SUSE/Redhat Enterprise Management tools provide enough to keep PCs uniform, clean, and correctly licensed. Centralized user folders make sense, of course. You can set power management settings via Group Policy, thereby theoretically lowering your power usage.
 
SpaceHonkey said:
Anyone here do disaster recovery? VDI is a no brainer in that case.
Click to expand...
I think that would highly depend on the disaster and the recovery implementation. I don't really see that VDI as a distinct advantage in DR.

It's my understanding that the biggest issue with VDI isn't actually the bandwidth requirement but the network latency. If that is indeed the case then streaming desktops over WAN could be too laggy to make an offsite backup and recovery via VDI feasible.

This is considering that your onsite servers are taken out by the disaster and can't provide the desktops.;
 
I look at VDI like this.

It is terminal server on a 1:1 scale that offers the advantage of remote 3D effects and environment enhancements like smooth flash animations / video playback or Windows Aero. If you already have / want terminal server or Citrix XenDesktop in your environment then VDI might be a solution for you, however if you are looking to move from an all desktop oriented environment to an all thin client environment you are going to see that there are huge costs in retraining your users, redeploying hardware, and the new learning curve behind maintaining a VDI environment.

From an MSP standpoint my company usually sells VDI to clients that have very old hardware (5+ years) and need an upgrade, but can't afford it. This works well for them because it allows them to extend the life of their old hardware, but be able to use the newest versions of software that normally couldn't run on their machine. We have also deployed it to a couple larger businesses in the area as a local instead of a cloud based service. They found that while they liked the management abilities, in the long run they only ended up converting their TS users to it, as their power users had demands that far exceeded what the environment could support.

VDI is not cheap, and as was said earlier there is no scale in which it is cheaper then just deploying normal desktops / laptops. Your support costs are the same, they just get shifted away from end user desktop support to server support. Your power consumption is the same as if you had just deployed laptops to everyone to begin with, if you end up buying thin clients you are averaging a per seat cost of $2,000 for new licencing and hardware (not counting server and storage). VDI is a more powerful terminal server, not a desktop replacement.
 
Latency is a killer for me. We run some Citrix XenApp farms for about 200 users across the globe and latency is the biggest issue we run into. riverbed, WanScaler etc won't help in this case either. Latency is a problem that I can't fix :(

oh and printing... what a kerfuffle

The cost is over the top and the performance isn't fantastic, its not really worth it.
 
Last edited:
SpaceHonkey said:
Anyone here do disaster recovery? VDI is a no brainer in that case.
Click to expand...
I work in disaster recovery (this weekend and yesterday were crazy thanks to the hurricane), we use VDI internally and offer it as a service to our clients. You you will never hear me call VDI a no brainer.
 
You must log in or register to reply here.
Back
Top