Virtualization I get it (I even love it) but VDI.... (Eric S. Perkins)

With the current state of our server operating systems virtualization is critical. No question. VDI on the other hand is a genius marketing product from VMware to gain market share from Citrix and Microsoft, and you know what? Citrix had to dump a ton of cash into developing a VDI solution just to handle the buzz! Kudos to VMware on this. VDI is bad choice 95% of the time.

Ok, first off I know I’m going to get a bunch of e-mails and comments about how I did not follow up my last piece with something directly relating to my opinions of the future of desktop operating systems, but this ties nicely and the entire series will have pretty much the same underpinnings in their thought process.

So, why do we have and need virtualization? Well, the obvious reasons right? It helps to segment server duties, delegate administration, make better use of hardware, and reduce datacenter footprints. I could go on all day on this topic, but the gist of this is that Virtualization is good and it is making IT better, no question. In fact, Virtualization has been described as a quantum shift in how IT works and I agree. There is one thing I may not agree on and that is that virtualization is necessary. I do not agree with this, I think Virtualization is a gigantic Band-Aid for server operating systems that were not designed from the ground up to be performing the duties they are performing, unlike the mainframes of yesteryear.

Don’t get me wrong; Virtualization holds many more benefits outside this like portability, superior DR capabilities, new ways to backup a corporate infrastructure again….. this is one of those lists I could go on and on about, but my point is that many (not all) of the things virtualization brings to the table are to circumvent or facilitate shortcomings in our operating systems. For example, some operating systems have great abilities to “jail” processes, others have great abilities to segment administrative duties and ensure that if a critical process fails it does not do the old BSOD in windows and kill the whole box. Virtualization would exist one way or another, but it thrives because our operating systems cannot facilitate all the capabilities of running it on an abstraction layer or hypervisor.

This brings me to VDI. I’m going to come right out and say it: VDI is generally not suited for large scale deployments yet especially compared to TS or XenApp. Ok, ok… there are some niche cases for it, but they are niche, without question. The school lab niche- I get this one, the developer niche that needs to be able to install his or her own software over and over again- I get this one too. But where is the niche that VMware, Citrix and Microsoft are all looking for…. the niche that is a good idea to deploy to thousands of users. It makes absolutely zero sense to run the same Windows or Linux kernel in system memory thousands of times. Many vendors promise golden images and ease of patch management but this is simply not true with the current technology period. VDI - It’s a pain, not only that what happened to all the propaganda when selling Server virtualization and how it would lower power consumption and it was more “green”. So now how is running the windows kernel for three thousand users three thousand times “green”? It is not efficient, not to mention you still need an end point running something even if it is a thin client (most are not) to connect and these devices need patches and to be secure as well. I’m sorry but for 95% of what I see, terminal services, Citrix and the other hosted desktop (shared) or published applications just make a ton more sense.

No administrator in their right mind should listen to the propaganda about users being able to install their own applications, or customize their own environments. If you want the ability for a user to have more robust profiles you need to use something like AppSense environment manager (an amazing product by the way) and let them connect to a terminal server desktop. A few months ago I made a bold statement on Twitter to the VDI zealots out there (many are VDI zealots as it gets them in the door). I challenged them to a debate on this as to how it has better served a customer and produced greater cost savings than traditional Terminal Services or Citrix XenApp and only one person took me up on it an then decided he would not debate me publicly. VDI just does not make sense, better operating systems and hosted desktops (with a shared kernel and O/S files) makes sense.

I know this is not overly technical we can dive deeper as we get into a discussion!

	Guest on Thursday, October 29, 2009
	WOW!!! I can't wait to see what kick's off now!

Jon Wallace on Thursday, October 29, 2009

Hey Eric,

I think I agree with *some* of your statements...

Love it or hate it, virtual desktops are the future but I don't think that's what your saying. I agree that in many situations, for example a call center that a XenApp (or Terminal Services) desktop would suffice quite fine. There are other sites that have made similar statements such as Brian Madden, Doug Brown etc, all of which love VDI technology but with it's correct use.

What is amazing however is where Citrix stand in this space - they're in such a good position. Citrix, in my mind without question have played this correctly and have positioned their products with perfection.

They have XenApp, XenDesktop and soon to be XenClient. These 3 solutions pretty much solve any virtual desktop requirement you have regardless of how you want them to be delivered, hosted or anything else.

Citrix is without question showing that beyond all others, it has listened to the community and people like yourself and is willing to adapt their products to fill the voids.

This is an interesting post - I need to digest it a little further but will no doubt have more comments...

	Guest on Thursday, October 29, 2009
	Comment From Twitter - win2ksrv @windowsregistry Agree mostly. Access to corporate apps in prestine environment is what matters. VDI facilitates move to BYOPC model.

	Guest on Thursday, October 29, 2009
	Personally I think Eric is dead on here, our company gets into a lot of companies that want VDI but they really need XenApp or TS. VDI is all smoke and mirrors except in a rare cases Scottie -VCP

Eric S. Perkins on Thursday, October 29, 2009

Hey Jon -

I agree Citrix is positioned very well, but with that said they are only positioned well because VMware did such an awesome job at bringing VDI to the forefront. Don't get me wrong I think it is total crap, but they are a machine!!!! :)

VDI should be used but in such small instances, you would need a XenApp type technology to fill the void for the larger user pool.

One more point XenApp (and in some cases TS) fills the void for BYOPC just as well. Publish some apps, or let users connect to a desktop, hell let them change the background to the family dog, I don't care. Running another entire bloated windows operating system just does not make sense and it's not "Green".

-ESP

Guest on Thursday, October 29, 2009

In your article you quote Appsense which I agree is an excellent product. IMHO it is essential that these profile management vendors create software to fit VDI as a solution instead of VDI as a product.

One of the things I like about Appsense is that it works on XenApp and XenDesktop in exactly the same way and although RTO and Tricerat do too, Appsense is just that little bit elegant.

This is an excellent post Eric and although I can see your view I don't agree with you in most. VDI will work, it will be mainstream and it will be used, why? Marketing plain and simple! VDI has some of the industries biggest marketing machines behind it and whether the technology is right or not, they will make it happen...

Josh Perik

	Guest on Thursday, October 29, 2009
	Comment From Twitter - roidude: RT @windowsregistry: Virtualization I get it (I even love it) but VDI....WTF? - http://bit.ly/3bNkQJ- (expand)>IT decision-making often irrational

	Eric S. Perkins on Thursday, October 29, 2009
	Josh- I agree with you man, the marketing will win.

	Guest on Thursday, October 29, 2009
	Eric, I gotta say I'm with you on this. Cindy

Michael Mills on Thursday, October 29, 2009

My 2 cents....

VDI as it is marketed by VMware is what you are describing and I couldn’t agree more. VDI as a point solution is largely overrated. However VDI has evolved to encompass much more than just "virtual desktops" in fact VDI is a strategy more than a product. To your point when organizations do their due diligence and analyze their users desktop and application needs, what comes up 90% of the time is that their users require nothing more than what Citrix has been doing for years, XenApp, Shared Session, Terminal Services, or whatever you would like to call it. And the other 10% of the time they will need a true isolated virtual desktop. To Jon's point, Citrix has really nailed this one, because above all else they understand that businesses are dynamic and so are their desktops and application needs. They provide organizations a choice, Hosted, Streamed, Provisioned, Published, Shared or delivered on a silver platter, it simply doesn’t matter, they provide the framework that you describe. (ok...that sounded a bit biased!!).

At the end of the day we all have to thank server virtualization for making the name virtualization synonymous with increased ROI and decreased TCO. Now, as leaders in the industry it is our responsibility to educate our customers on the best possible solution.
Great Article..!!!

	Eric S. Perkins on Thursday, October 29, 2009
	Nice! Thank Mr. Mills! Hope you are well my friend.

	Guest on Thursday, October 29, 2009
	Eric is right, I agree the marketing machines are in full spin mode

	Guest on Thursday, October 29, 2009
	Comment From Twitter - lynxbat: RT @windowsregistry: Virtualization I get it (I even love it) but VDI....WTF? - http://bit.ly/3bNkQJ (expand) < what about instantiation & mgmt??

	Guest on Thursday, October 29, 2009
	Did we miss the purpose? VDI is more flexible alternative to buying an equal set of hardware + management + people to manage and deploy all this hardware. It's very green in that regard. Are there other ways to avoid buying all that hardware? Yes, many ways to skin this cat. Rob Markovic @vRobM

Jon Wallace on Thursday, October 29, 2009

Personally I think people miss the point when it comes to 'green' and VDI. Remeber guys that even though you have a 'virtual' desktop, you still need a 'physical' device of some form to access it. Granted that these devices are thin-clients but have you seen today's thin-clients, they're as big a PC's in a lot of cases. You still have a monitor at the client, keyboard, mouse and enough plastic for those devices to cause a substantial carbon footprint.

Yes, the lack of need to throw said devices away after 5 years is not there but that can be negated with recycling anyway...

For me, VDI's huge benifits are around centralization, management and the removal of the desktop-refresh need. VDI is benificial but XenApp will always have a place in the workplace, regardless of how good client technology gets!!!

Eric S. Perkins on Thursday, October 29, 2009

Rob- Yes I think you are missing something. My point is that the alternatives to VDI like Terminal Services and of Course XenApp are 95% of the time a better choice. VDi is no where near as "Green" as SBC (Sever Based Computing) simply from a user density perspective.

Also while thin clients "may" and I stress may use less power you are still using power and those device still need management. It's marketing fluff. But you are also correct in saying many ways to skin this cat, I just tend to think VDI is not a good move unless you are in a ver specific situation like I outlined in the article, and yes there are a couple more.

Jon- the points you outline for VDI apply perfectly well to SBC, and the management is less of a headache.

	Guest on Thursday, October 29, 2009
	Comment From Twitter - djfeller RT @windowsregistry: Virt I get it but VDI..WTF? - http://bit.ly/3bNkQJ < wht abot other desk virt flavors beynd #vdi http://bit.ly/4bfED7

Eric S. Perkins on Thursday, October 29, 2009

Daniel - Nice article I think we agree more than we disagree. at it's most basic level I'm opposed to running multiple kernel's and operating system resources to facilitate users access to what they really care about. Users care about the environment they use, and quick, easy access to line of business applications to complete a job function. Many SBC solutions already do this quite well, by adding in something like an AppSense solutions you can get very close to a desktop experience with better security and ease administration, patching etc...

Going back to my last article, I think you are going to see more and more "Business" applications moving to the web browser like Oracle, Salesforce etc, with power of Microsoft's silverlight platform, flash and other delivery mechanisms Virtualizing an entire desktop operating system just does not make sense (most of the time). This is a topic for another article as I don't want to get off track on this discussion.

One thing I did not address in this article as I think it is a Pandora's box is application virtualization which I am heavily in favor of. This allows admins to easily deploy apps to whatever environment they are running SBC, VDI or thick client.

Guest on Friday, October 30, 2009

I agree with you Eric. Maybe what is needed rather than VDI is for Microsoft to make the Terminal Services environment more like the desktop environment. I know it has made great steeps in the last few years but a users persepective is that it's still not the same as what they are used too. If it looked and felt just like windows7 then it would be up to Citrix et al to workout how to get this to the client efficently. For Citrix to really master this IMHO they need to fill the one hole in ICA which is 32bit colour.

Benedict Heaver

	Guest on Friday, October 30, 2009
	"I think Virtualization is a gigantic Band-Aid for server operating systems that were not designed from the ground up to be performing the duties they are performing" Amen! And you might add "crappy applications" to that.

	Guest on Friday, October 30, 2009
	While I agree with some of this article, Citrix always preformed so poorly in the past it just never worked well for us. Nice Article - Mark

Eric S. Perkins on Friday, October 30, 2009

Hi Benedict - Nice comment! Windows 7 is a huge step forward for Microsoft no question, but I'm not sure this is a good idea to host a Windows 7 type environment anytime soon (maybe as one silo) at a large corporation as users just can not handle change :)

I think windows 7 is going to gain widespread adoption in corporate environments unlike it's redheaded step child brother Vista. This is how a product like AppSense can help as you can manage the users environment on a hosted (Shared) desktop environment like XenApp or TS enabling easy profile migration to new environments as they become available. So if Server 2010 or whatever is next offers the Win7 interface you could easily migrate those users settings. I did this often with the profile changes from Server 2000/2003 to 2008.

As for 32-Bit color the RDP client supports 32-Bit Color, Has Citrix done anything with all the hype around HDX to support greater color depth? I have not had time to research ICA this summer as I have been focusing on long-term IT road mapping with customers.

Also on this topic it is well known I do not think the desktop operating system will matter in the next 5 years. Servers will matter, the web browser will matter applications delivered to cross platform technologies like HTML, Flash, and Silverlight will matter. Legacy applications (difficult apps to update, or expensive) will be relegated to SBC (TS or XenApp)

Guest (two posts up) Thank you for highlighting that!!! “"I think Virtualization is a gigantic Band-Aid for server operating systems that were not designed from the ground up to be performing the duties they are performing" Amen! And you might add "crappy applications" to that.”

I wish you would have put your name, as I agree with you completely I should have added “crappy applications” :)

Guest on Friday, October 30, 2009

Eric, good post, and we agree that VDI, by itself, doesn't deliver the cost savings and operational efficiencies IT needs, or the more productive desktop experience end users want. But VDI and client-side hpervisors do make possible new approaches to desktop management that can holistically address image management, application delivery, patching, desktop repair, personalization, asset management, data protection, disaster recovery, and more in ways that are simply not possible with physical desktops and first-generation PC lifecycle management products. I'd welcome the opportunity to show you what we're doing at Unidesk (www.unidesk.com), and get your feedback.

-Tom @tomrose_

Eric S. Perkins on Friday, October 30, 2009

Hi Tom - Thanks for the post. Conversations like this are very helpful! I guess it depends on how you look at it. I agree 100% that hardware agnostic operating systems (Like what a client side hypervisor can provide) (or a better operating system design) is a good thing for IT. As you point out the many advantages.

My contention is the marketing around hosting desktops as a replacement for SBC is misguided. Actually I will take this a step further I challenge any CTO/CIO that has decided VDI is the way to go for the masses to a debate on this! Public forum, and the winner buys dinner in Paris! :)

Client Side I get, and agree with especially while the desktop operating system piece is so critical, but hosted out side of the niche examples I give above just does not make good sense from an ROI perspective, from an efficiency perspective, or from an administrative perspective.

Call me anytime to check out unidesk in more depth, my contact info is on my webpage, just click the chimerically logo above. I have checked out unidesk before but not in any depth.

-ESP

Guest on Friday, October 30, 2009

Someone mentioned this follows the BYOPC model. BYOPC is also a new stupid buzzword that was probably created to help justify the idea of VDI. If I heard my employer say BYOPC I would tell them to get bent. It's their responsibility to provide me with tools to perform my job, not the other way around. VDI is stupid, BYOPC is stupid, and I hope this FAD goes away quickly. XenApp has over a decade of a proven technical worth, VDI has a long, LONG way to go and I hope it never gets there.

	Eric S. Perkins on Friday, October 30, 2009
	Guest - I agree with you 100%

Guest on Friday, October 30, 2009

Eric nice post. VDI is not for everyone just like some power/heavy/IT users will need more than just SBC/XenApp/Terminal Services. I see more value with a combination of application and desktop virtualization where majority of the users are using application virtualization being XenApp/Terminal Services/ThinApp/App-V or whatever while the power/heavy/IT users use destkop virtualization - XenDesktop/View or whatever. While client side hypervisors and offline desktops gives desktop virtualization more value. I compelely agree with the comments about management products like AppSense, etc give more value whether you are using application or desktop virtualization. The VM Hosted Apps features in XenApp FP2 does give SBC/XenApp more value but even with that feature you will still have virtualized desktops somewhere whether you are using SBC or VDI. Just my two cents.

jarian @jariangibson

Eric S. Perkins on Friday, October 30, 2009

Jarian -

That is a scenario that works, agree, the problem is it over complicates IT. While many organizations are trying to streamline IT and cut costs running a large combination of XA, XD, Streamed apps, locally installed apps, Thick clients, and eventually client side hypervisors it just seems like we are trying to keep ourselves busy!

VDI makes sense, a lot of sense to VMware!!! as they needed this technology to make inroads to the Citrix customer base, I truly think this was an incredible marketing move, it even made Citrix develop!

Thanks for your comments man!

	Guest on Friday, October 30, 2009
	Eric - I am with you on this one as well... -- Greg Jackson

	Guest on Saturday, October 31, 2009
	Great article and discussion. Thanks for your insights Eric on the future of Virtualization vs VDI ....VMware....

	Guest on Saturday, October 31, 2009
	Comment From Twitter - mrtomahawk @windowsregistry tottally agree.... and many the small niche things which you can't do in TS/RDS is getting smaller, and smaller over time

Guest on Sunday, November 01, 2009

While I have a lot of respect for you Eric, I believe, and these are purely my thoughts only, you might be wrong. This is primarily based on my experience over the past 13 yrs as a TS / SBC (Citrix Pres Server) / VDI consultant.

The kernel is just one piece of the puzzle. Green IT is another, and thin clients or in fact some form of client are required for terminal services / citrix, in most cases traditional SBC works best via a PC client and not a thin client...so much for green IT there. I agree with your kernel point, its not the most efficient way to load up a system for 3000 users, but that is just one technical component to an entire solution.

VDI can work very well when its architected the right way. SBC can work very well also, and both have pros and cons. Getting some legacy apps working with a Windows Server OS can bring alot of pain, and those people who have worked in many environments know that you need some smart app delivery people to get some apps to work on windows server that were only ever tested by developers on Windows XP for example. This is one area that VDI excels. And it sometimes makes such a difference, that customers will pay that bit extra just to show their users a familiar Windows XP look and feel. I think SBC has burned a few customers in the past, I worked for a large bank where it took 12 months to recruit 13 Citrix experts, because a) they were hard resources to find, and b) they all wanted extremely high salaries. VDI has this problem to a degree, but its a lesser extent in my opinion, and most folks (IT support) know how to fix common XP problems. But it takes a different calibre of personnel to be able to support a large Citrix farm. And this costs $$$. Operational expenses, which are usually 3x the cost of upfront Capital Expense.

VDI brings its own issues, but it is more than a technical arguement. The CIOs & CFOs that sign-off on SBC and VDI projects don't need to get down to the deep technical layers, they just want a solution that delivers results. If that means a familiar desktop experience at a higher cost... they might just sign-off on that. I have seen it alot in the last 3 years, and in the last 12 months I have been surprised by the uptake and satisfaction with VDI. Maybe its marketing thats driving customers towards something new, but if it wasnt a solution customers wanted, why is there multiple broker vendors? beyond Citrix and VMware. Infact the other vendors had products and sold well for months and years before Citrix and VMware ever released a "VDI" product.

You might not have seen the uptake I have... but trust me... it's here to stay for a while, and Windows 7 will be the catalyst for customers to try something new, some will stay with Citrix, some will stay on full PCs, but there will be many that shift to VDI, as a way to deliver a better solution for the end-user. Thats really all we are all in a job to do... isn't it ?

Cheers.

Eric S. Perkins on Sunday, November 01, 2009

Hi Guest!

I think we agree on many points. Thank you for your comment. Looking forward I have some long term thoughts on this and I do not think VDI is here to stay. Some companies will adopt it sure, some companies might even do large rollouts of it, but it is not efficient in any perceptive $$ or operational efficiency (my opinion of course).

I guess the big thing I don't agree with you on is that applications can be difficult to get to work in SBC environments. To be honest this is a problem of the past. I'm not saying it does not exists, but I just don't see it very often anymore, and if the problem does exist and it just can not be overcome, I could not see a large rollout of VDI because of one poorly coded app that won't work in SBC (again a problem I just don't see anymore).

As for your point on multiple connection brokers, one word- "Bandwagon". VDI has tremendous buzz around it, and companies are making money on small deployments and lots or POC's as a small company (or not) it's easier to position a product that has marketing momentum from heavy hitters like VMware.

Long term my personal opinion is that the desktop Operating systems in corporate environments will be de-emphasized (3-5 years), the de-facto client will not be RDP/ICA/other, it will be a web browser. This is clearly a different conversation, but keep your eye on companies developing serious Line of Business Applications through the browser (think Oracle).

Again nice comment! Thanks!

-ESP

P.S. I will help my customers deliver VDI if that is a decision they have made! :)

Guest on Monday, November 02, 2009

In total agreement with Eric here - but the train has already rolled out of the station!

Ultimately VDI/HVD as a wholesale deployment idea will fade away as people think about the added complexity and cost being introduced their environment vs any supposed benefit - as Eric has said it works for selected use cases - for everyone else TS based solutions and application virtualisation are the most suitable technologies.
However the hype and focus on VDI has now driven a ton of innovation and development that otherwise may not have happened - as a result new ways of working are being uncovered everyday - thats got to be a positive thing!

Tyrone Thomas

Guest on Monday, November 02, 2009

Hi Eric,

Thanks for the post. My name is Victor Thu with the Citrix desktop virtualization team. I am responding to some of your concerns and you can find my response in Citrix blog post here:
http://community.citrix.com/pages/viewpage.action?pageId=97715047

In addition, I’m extending an offer to you to meet with our team. We can share some valuable insights on this topic.

Best Regards,
Victor Thu
Senior Product Marketing Manager
Citrix Systems, Inc.
http://www.citrix.com/xendesktop

Guest on Monday, November 02, 2009

This article is the equivelent of Fox News, it's not news it's opinio-tainment. Stating that "VDI is a bad choice 95% of the time" in the first paragraph of your article which in case you did not know, 100% of the readers will read along with the last paragraph and not what's in between, is not a wise decision. Calling it "stupid" without understanding what the requirements are for the thousands of businesses worldwide seals the deal. VDI is a problem resolution for many organizations that had no other reasonable solution options including Citrix Metaframe ($$$$+incompatibilities) and MS Terminal Service (crash / fail). VDI was the ultimate focus of VMworld this year with a ton of vendors now preparing a full product suite for its various implementations, but hey, they must have been stupid too, right?
A properly designed and managed VDI solution is the foundation of tomorrow's corporate desktop and if it isn't then where have all these other grand solutions been all these years?

Guest on Monday, November 02, 2009

Hi Guest (last post).

Well you are wrong. And my fellow Citrix and Terminal server admins will back me up, incompatibilities are almost non existent nowadays this is marketing propaganda that just is not true any longer. Sure one or two troubled apps could exist but that does not justify running a desktop kernel in memory and a bloated O/S for every users makes perfect sense.

Just because a ton of vendors are preparing product offerings does not make it right? If I was a software company I would do the same thing. They are capitalizing on industry buzz. And how do you know I don't understand the requirements of many corporations?

My competitors, peers all walk in the door because VDI get's them in and 95% of the time they walk out having proven TS/XenApp is a better choice except for the niche plays like I outlined above.

Eric S. Perkins on Monday, November 02, 2009

Hi Victor,

To help frame this better as every vendor has a different definition for VDI and the related technologies I am addressing VDI in the sense of a hosted desktop separate virtual instance per user running a full desktop operating systems in the datacenter.

Much of your response I agree with, but simply because it strays away from the above context. Here are some points I agree with you on.

1. Local streamed desktops, I'm a big fan of this technology it's not quite ready yet but you guys are getting very close! Citrix is the clear leader here no question.

2. Hosted Shared desktop we have been doing for years, again I'm a big fan have been since Winframe. What is not to love, it's "green'er or ish" easy to manage, patch etc etc... But I have not seen this called VDI before this is just classic TS or XenAPP type technology.

3. Streamed or Virtual apps to any point SBC, Local Desktop etc.... makes a ton of sense again I have not seen anyone else call this VDI. But I agree makes a lot of sense.

My contention is nobody has the patching, the golden image thing, or the management truly down yet. It makes no sense to run a windows kernel 3000 times in a datacenter for 3000 users when shared TS environment will work 95% of the time. While some may say that this stuff is being worked on and they are probably right, maybe it will even be right at the end of the day it still makes no sense to launch a full O/S per user and support a thin client or other end point. Application virtualization makes sense, sever virtualization makes sense, SBC makes sense, revisiting legacy code and producing/updating applications to run in a web browser, flash or silverlight makes sense.

Eric

	Guest on Monday, November 02, 2009
	Completely agree and have ever since VDI popped up. A solution looking for a problem to solve. I'd much prefer vendors to fix their software to run correctly under XenApp than VDI. Rob.

	Guest on Monday, November 02, 2009
	>I truly think this was an incredible marketing move, it even made Citrix develop! That in itself has been a very good thing IMHO. Citrix were sitting on their laurels too much IMHO. Rob.

	Eric S. Perkins on Monday, November 02, 2009
	Rob... YES and YES. :)

	Guest on Monday, November 02, 2009
	Is the offer to Publicly debate VDI viability still on the table? If so, i would like to accept the challenge -T.Rex Rohrer Liquidware Labs Inc. trohrer@liquidwarelabs.com

Eric S. Perkins on Monday, November 02, 2009

T.Rex - Absolutely, We could see if Briforum for next year has a slot open he would let us use. actually that would be in your favor as by then VDI might actually work better by then! :)

So we would have to set some metrics as to what we are trying to prove and disprove

ex. Administration, Application deployment, Storage requirements, Power utilization per user, licensing costs per user, this kind of stuff right? My Preference would be to have a neutral third party prepare the points like Ron Oglesby, Jon Wallace, Doug Brown or Brian Madden or other......

The other thing that would be difficult but necessary is that we need to use the same testing metrics as we can both make numbers say anything.

And to be clear we are talking about Desktop Operating systems running with a connection broker not the new invented Citrix definition right?

Another idea would be we could draw up a set of Visio Diagrams with supporting documentation as to licensing costs, and capacity per server and submit them online for the community to vote as to what they would put in there enterprise, this might be easier to do and would have the same net effect while allowing us to get this done pretty quick.... Use a set base of LOB applications like Oracle apps, MS Office, and a some random app.

Actually I like this better, and then if it works we can always present them in person at the next large forum?

That would be fun!

Guest on Monday, November 02, 2009

Eric I think you need to also discuss why people centralize. Get's get that out of the way first. Tons of benefits blogged about many times. Then the debate is how. Desktop single instance OS or Shared TS OS. Desktop OS offers one way to do this for numerous benefits but there is a cost and benefits like one OS to support across all your desktop use cases, app compat perception/vendor support may be real may be fud. Is it niche for the vast majority? I think that will depend on costs of supporting multiple OS's in the data center, impact of client virtualization and which OSs to put where etc. For me.

1) Central desktops mean more security
2) Supporting one OS is easier, and desktop OS helps reduce complexity
3) TS is cheaper, but you loose flexibility. Sure you can do 1-1, but again it's a server OS license now vs. VECD.

So not understanding centralization is STUPID. MOST PEOPLE ARE STUPID, MANAGERS who make decision and the STUPIDEST and are the ones that buy VMWare desktops for reasons they can't really explain.

	Eric S. Perkins on Monday, November 02, 2009
	Sure all good points, except your supporting on OS with TS or XenApp as well. What flexability do you lose, and is it flexibility that matters?? Thank you for your comment! -ESP

Guest on Tuesday, November 03, 2009

Hi Eric, great article because it opens to discussions and debates.
Is VDI BS? Is SBC that crap? Is Desktop Virtualization only a big marketing buzz?

I leave your readers answer those questions...

Let me add my thoughts here if you don't mind. I search your article for the word 'security' and I don't find any! So I'm asking myself how do you guys live and sleep at night knowing that your company's most private data could sit on a lappy out in the wide?

Is the concept of security gone over the last 2-3 years? Did Microsoft OS'es become flawless all of sudden? Don't you have to patch, upgrade and patch again all those desktops/laptops anymore?

On average a workstation is being used 1900 out of 8760 hours a year. The rest of the time (80%) is it idle!. Companies can save a lot of money by consolidating and this is all about.

To conclude I would like to post a link to another very interesting article: http://www.cio.com/article/504348/Desktop_Virtualization_5_Most_Popular_Flavors_Explained

We've been doing Desktop Virtualization for a long time now, and VDI a la Citrix or VMware is just an 'enhanced' version of it...

Rgds,
Didier Pironet

Eric S. Perkins on Tuesday, November 03, 2009

Hi Didier-

Thank you for the post. I take security very serious, just only so much content for discussion at one time! :)

I prefer my companies, customers and everyone's data resides in the datacenter, so to your point I'm not a big fan of locally installed LOB apps or people running around with spreadsheets full of social security numbers! A step further I also think Identity theft should qualify for the death penalty!

As a SBC proponent I would not advocate laptops running around with data, but I have yet to find a large enterprise that does not need this in one way or another (connectivity etc...) so we rely on streaming apps, encryption and all the other stuff you would think like two factor, disk encryption etc...

Please excuse typo's, no coffee yet!

Eric

Eric S. Perkins on Tuesday, November 03, 2009

Oh, one more point SBC is the right solution most of the time NOT CRAP! :)

VDI is marketing Buzz (and in general crap) although Citrix wants to lump it all together Applications streaming and profiles are a whole different thing (one that I agree with Citrix is an important part of the puzzle) but necessary with whatever you do SBC, VDI, Thick, Thin all of it!

VDI is a waste of resources (no user densities compared to SBC), I can give my company/clients/users, a great environment complete it with AppSense and even let them put the family doggy as the background picture. I don't want my users installing apps anyway (except in the cases cited)

Ok now coffee!

Guest on Tuesday, November 03, 2009

Post from Appdetective Eric to answer my earlier post of why you loose flexibility with a TS OS, let me add a few point, and challenge your assertion that VDI is just crap.

- The problem with shared OS, is exactly that, it's a shared os. That logically means you do not control the full OS, so x,y,z things. It means I have to build software packages and test for my fat XP environment and SBC environment. This adds to software lifecycle costs at scale.
- There are some vendor support issue, although with R2 I think in time if the world moves to x64 this goes away largely.
- So you may ask why not 1-1 XA. Sure you can do that and pay for a server license as opposed to VECD so it's like a desktop. I will say however it reduces my package testing issue.
- What people miss IMHO is that nobody sits on my lap when I use a desktop and types at the same time. I am trying to deliver a desktop service, that is able to be connected to from anywhere, on any device. That comes at a price for full flexibility.
- The other problem with both Xd and XA is the broker. Those things go down and there is broad impact that I can not accept in my business. There the need for a HDX Connect type feature. It's a about the reliability of a desktop.
- This type of design lets me install what ever I want on a desktop OS. I get over silo's of SBC etc, and BCP is built into what I do.
- So key point here is capital cost upfront if is > on VDI vs. SBC. However you need to understand what service you are trying to deliver. If you are trying to deliver a true desktop service then SBC is CRAP! Shared infrastructure, reliability, crappy apps, complexity etc.
- That you want to deliver a lot of the same, SBC is just the same. Appsense is great but complex.

I love the fact that Citrix now offers multiple delivery models under one license. To simply say one model is CRAP is just naive. Upfront cost is often fine when the greater goal is complete business agility.

Good debate though, and I think it comes down to people to asking the question why.I think this blog at Citrix helps explain it well. http://community.citrix.com/display/ocb/2009/10/06/Mending+broken+hearts+with+XenDesktop+4

Guest on Tuesday, November 03, 2009

My two penny's worth.

There is not going to be one single technology that is right for all users. Currently we have some users who are a good fit for SBC and others who get a laptop or traditional desktop. Going forward, some users, mainly traditional desktops, will be a good fit for VDI (if they are fine on XenApp, why move them). Others will not. Longer term, a proportion of laptop users will be a good fit for client virtualization.

I do not see a silver bullet technology that would optimal for all users - it all comes down to choosing the right mix of technologies for the particular circumstances.

Martin Ingram (AppSense)

	Guest on Tuesday, November 03, 2009
	DEAL! BriForum it is - The great VDI DEBATE I suggest we create a panel. Any other takers? T.Rex trohrer@liquidwarelabs.com

	Jon Wallace on Tuesday, November 03, 2009
	I would imagine the people you certainly want as part of this are Eric, Brian, Martin, Doug, Simon Crosby and maybe a couple of ther names... I think the format should be in the form of the Presedential Election Question Time where by each delegate gets a strict time period and 1 or 2 chances to rebuttal... Just my thoughts anyway but would certainly like a VDI showdown like this...

	Guest on Tuesday, November 03, 2009
	I tend to agree with some of the posters above, that this is about understanding use cases and service levels and understanding what why you are delivering something. Harry Labana (Citrix)

	Eric S. Perkins on Wednesday, November 04, 2009
	"Jon Wallace on Tuesday, November 03, 2009 I would imagine the people you certainly want as part of this are Eric, T.Rex, Brian, Martin, Doug, Simon Crosby and maybe a couple of ther names..." Shawn Bass threw his name in the hat on Twitter, for sure a good addition. I'm traveling today but had something I wanted to add, I'll get it up this afternoon. -ESP

Eric S. Perkins on Monday, November 16, 2009

Busy week, sorry for not getting back to this sooner. First off, thank you all for joining me in my debate! It’s been great.

Harry Labana (Citrix): Let me talk to your point first, OF COURSE! But the problem with VDI is these cases are so small in scope it does not make sense for large scale deployments. As I wrote in the article:

I get the niche cases.

Dev shop in India
School lab or computer lab
Contractors needing an environment to work in but secured

I even concede that there may still be some apps that you cannot get to work in Citrix and TS environments, but this list is pretty small and to be honest, I have not run across it in enterprise environments in some time. VDI consumes more disk space in all cases, and more server resources in all cases.

I know these are being actively worked on and constantly improved but why? Why not throw those dev resources at improving SBC which is a superior thought process, and more efficient than running an O/S kernel thousands of times.... while application programmers should be putting resources behind web enabling applications (private or public)(i.e. Silverlight, AIR/FLEX, HTML).

Martin Ingram (AppSense) - I know somehow client side virtualization got added into this debate further up in the comments, but again I must say this is marketing spin. Citrix needs to position themselves superior to VMware (and don’t get me wrong - in this space I think they are) so they took the term “VDI” and added other technologies to it not directly relating to the term’s initial definition, actually it’s genius.

My article or thought process is specifically in regards to VMware and Citrix (and other smaller players) trying to convince CIO’s and CTO’s that it is a good idea to host desktop operating systems in the thousands at the datacenter instead of SBC.

IMHO client side virtualization is an entirely different conversation. Speaking off the cuff, I actually like the idea in some business cases of a client side hypervisor and streaming down a business specific OS complete with apps etc... I really don’t want to get into this topic on this article though.

If the users you mention that are currently on traditional desktops are candidates, would they not be better candidates on SBC? Throw some AppSense on the servers, lock them down, and still let them change the wall paper to their kitten at home or whatever makes them happy. I simply do not ever talk to CIO’s or CTO’s at large enterprises that say to me “Eric, ya know we want our users to install apps, and cause our helpdesk to go crazy rolling back snapshots of VDI images...”

Last one for this morning! In this case I would like to go point by point from Appdetective:

Appdetective - The problem with shared OS, is exactly that, it's a shared os. That logically means you do not control the full OS, so x,y,z things. It means I have to build software packages and test for my fat XP environment and SBC environment. This adds to software lifecycle costs at scale.

ESP - So what you are saying is that you would build images for your FAT XP environment and use those same packages on your VDI environment. I doubt it. Just because VDI has a separate user space for every user does not mean resources are always plentiful. You would still want to build those images for the VDI environment. Another answer to this is to host specific apps only on SBC.

Appdetective - There are some vendor support issues, although with R2 I think in time, if the world moves to x64 this goes away largely.

ESP - We are going to have to agree to disagree on this point as I have not found a modern enterprise LOB app recently that does not work very well in an SBC environment.

Appdetective - So you may ask why not 1-1 XA. Sure you can do that and pay for a server license as opposed to VECD so it's like a desktop. I will say however it reduces my package testing issue.

ESP - I would not ask this question! ;-) it defeats my entire argument that it makes no sense to run a bloated windows operating system thousands of times.

Appdetective - What people miss IMHO is that nobody sits on my lap when I use a desktop and types at the same time. I am trying to deliver a desktop service, that is able to be connected to from anywhere, on any device. That comes at a price for full flexibility.

ESP - Flexibility has a price, giving admin rights to a user has a price, it’s called a massive help desk. Everyone keeps talking about user experience in the VDI debate. IMHO this needs to be defined better. We need to define wants and needs. Needs are things like, application performance, access from anywhere, datacenter fail over capability, stability. Wants are things like changing the mouse pointer to the dinosaur icon. User experience needs to be looked at from an ROI perspective only, can the user get their job done quickly, efficiently and consistently while not causing storage needs to go through the roof and adding tons of computing resources to support it. I don’t think anyone here will dare debate that VDI consumes less resources.
Appdetective - The other problem with both Xd and XA is the broker. Those things go down and there is broad impact that I can not accept in my business. There the need for a HDX Connect type feature. It's a about the reliability of a desktop.

ESP - I think you bring some really great points to the discussion but this one I don’t agree with. Build the environment better. My Citrix/TS environments do not go down and to stay on topic I’m not sure how VDI (datacenter hosted desktop OS’s) would make this any better, actually worse I would think, especially given the maturity of the offerings compared to SBC.

I have to run this morning, tons to get caught up on but thanks for all the comments this is a great discussion, I’m just trying to wrap my brain around the validity of VDI and so far I stand by my statement that it’s marketing crap.

Eric S. Perkins on Thursday, November 19, 2009

Just a clarifying statement to help the influx of e-mail on this topic!

I am ONLY including VDI in the sense of Servers hosting desktop operating systems with a connection broker.

I think the other solutions Citrix is trying to package into the term VDI like Provisioning, Application Streaming, Application Virtualization (other), Client side hypervisors etc etc etc etc are all very interesting and I think everything has it's place. My point of contention is that VDI (Traditional) does not make sense in large deployments VS. other options like SBC.

	Eric S. Perkins on Friday, December 04, 2009
	Just completed a podcast with @micheal_keen & @djfeller on this topic. I continue to be excited about the developments in the industry no matter what it's application. VMware, Citrix and Microsoft are fueling a competition that at the end of the day will benefit all of us.

	Guest on Friday, December 04, 2009
	Do you have a URL to the podcast?

	Eric S. Perkins on Friday, December 04, 2009
	I will post it as soon as Dan Feller from Citrix gets it on the his site.

	Eric S. Perkins on Friday, December 04, 2009
	sorry typing from mobile...

	Eric S. Perkins on Monday, December 07, 2009
	Here is the link to the podcast: http://bit.ly/6Fsce6