Disappointed! (the true sad story of client hypervisor meloncholy) - Tim Mangan - BrianMadden.com
Brian Madden Logo
Your independent source for desktop virtualization, consumerization, and enterprise mobility management.
Tim Mangan's Blog

News

  • Two App-V Books "The Client Book" and OSD e-book Reference at http://www.tmurgent.com/Books/default.aspx. Training classes now booking for App-V 4.6 SP1.

Past Articles

Disappointed! (the true sad story of client hypervisor meloncholy)

Written on Jan 10 2011
Filed under: ,
6,791 views, 8 comments


by Tim Mangan

Lump of Coal
Attribution: OpenPhoto.net

All I wanted for Christmas was a decent client hypervisor.   Two years in a row now.  Was that really too much to ask for?  Last year at Christmas I was only a little disappointed because I had kept my expectations in check.  But surely by this year I would have one, right?  I was excited by the news that VMware announced an upcoming type-1 client hypervisor.  Citrix was working on one.  Virtual Computer had one already, but it was limited.  Microsoft had not made any announcements, but you had to assume that they had one in the works but decided there was not a good enough reason to make it available yet.  I was sure that 2010 would be the year I get a decent client hypervisor for my laptop.  Surely I would have one for Christmas this year.

Dang.  I tried.  I threw a party to see who would show up.  VMware, which announced they were coming early on,  later sent their regrets.  Not much in the way of an explanation, but the gossip is that the whole HCL (Hardware Compatibility List) thing was just too much for them.  One suspects they are continuing to work on this behind the scenes, however, as long as nobody else has one, they can keep shipping their Type-2 client hypervisor which means that they are in no hurry.

Virtual Computer actually came to the party early.   Last summer I tried out a beta of their 3.0 client hypervisor.  They significantly expanded their HCL list to include Radeon graphics chipsets so I could run it on my latest high performing laptop.  No lame V-Pro limitations here.  The biggest hassle with installing a type-1 hypervisor on an existing laptop is the destructive install thing.  Fortunately I had thought ahead and bought a laptop with a second hard drive bay.  So I popped in another drive and went for it.  I soon found that I had to swap positions because they hadn’t considered that use case and it would only load the hypervisor on the first hard drive, but I was OK with that.  The Install was straightforward.  And it worked on my laptop.  First of all, having a hypervisor with a web browser built in is awesome!  Boot ultra-quick and get access to the internet without an OS.  Yeah, some people worry about the security of the hypervisor when you do that, but I’m not.  Those threats need a file system.  I also installed the server software and had it stream some VHDs to run.  It was quite impressive.  Talking to the folks at Virtual Computer, they seem to really feel that their value lies in the Management side, not the hypervisor.  And while that might be their value as a company, and for many enterprises, my interest is just in the hypervisor.  I really liked the NxTop.  The performance was very good. The things I didn’t like were nitpicks.  Being a linux based hypervisor, my touchpad was overly sensitive and even when toned down by the generic touchpad controls was not something I wanted to use on a daily basis.  The wireless network worked well, but I live in a home/office with two segregated wireless networks and I bounce between them a lot.  With this setup I had to jump from the OS I was running to get to the hypervisor to make the change.  I really wanted to do it from the virtual wireless adapter interface in the VM tray.  While these may be small things, a laptop is personal and these usability issues are significant. Ultimately, while I really liked it, I ended up reverting back to my old setup after a couple of weeks.

Citrix tried to make it to the party.  They  released XenClient 1.0, their type-1 client hypervisor, but Brian said don't waste your time because it wasn’t really ready yet.   I did anyway.  While Citrix has a small V-Pro only HCL, they did indicate that it should work, even if unsupported, on a wider range of hardware.  So I swapped in another hard drive (you didn’t think I left that second bay empty after buying the second hard drive did you?) and tried to install.  No luck even getting it to install.

So where am I now?  Just where I started.  Using the best type-1 client hypervisor out there today for my needs.  Hint, it is one not on the typical client hypervisor list. 

I know that my needs are not mainstream.  I’m not looking for something small and light so I can access VDI/Terminal Server/Cloud apps locked up somewhere else.  I don’t need management stuff (yet).  I just want to use the awesome power you can get in a laptop these days without being disabled by a “bad internet day” or a semi-stable wireless routers in the overly concrete basement of a certain swanky hotel in Chicago.   I want to bring my lab with me when I go.  I need machines to write software, machines to test software, machines to demo software, plus my personal machine.  I want the benefit of segmenting those needs into different OSs, with the ability to revert those test and demo machines.  I demand performance, but still demand that the touchpad and wireless work.   So my choice might not match yours, which I’m sure many of you will tell me in the comments!

I’m using Hyper-V on the laptop.  It is providing the best performance with all of the features I really need, except one.  There is no hibernation.  I survive without that.  I can “Save” (suspend in human terms) the individual VMs and only need to shutdown/boot the hypervisor and root partition.  But otherwise it allows me full use of the hardware I bought at a decent performance.  Yes, I still need to go to the parent partition to change the wireless router, but I guess it seems less of a burden than with Virtual Computer.  I’m waiting patiently for the RTM of 2008 R2 with SP1 (in an extremely unusual departure from my normal beliefs, I don’t want to install the RC on this one) so I can gain the advanced memory improvements. 

And since I love unconfirmed rumors I’ll just mention that a Hyper-V V3 (aka “Hyper-C”) from Microsoft would be a great present in 2011.  This Microsoft client hypervisor is being discussed  as being part of Windows 8, but if Microsoft thought it was in their interest to get a type 1 hypervisor for the laptops with a “MinWin” parent partition out sooner they would be able release it independently.  Heck, as far as I’m concerned they don’t even have to wait for Christmas next year.  There are a lot of holidays to choose from and I’m not picky!

Tim Mangan

 Author: Tim Mangan is a Microsoft MVP for App-V, a Citrix Technology Professional, and holds the position of Kahuna at TMurgent Technologies. He runs a local virtualization user group in Boston and has spoken at every BriForum, as well as other venues. Read more at his home blog or website.

 
 




Our Books


Comments

appdetective wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 9:01 AM Link To This Comment

The client hypervisor is obviously very hard to get right and there will be many more years of evolution only to be given away for free. So what is the business model? Management and security play for sure, but Citrix is doing nothing much there that I know of that would keep them ahead of MS with a SCCM story for long if MS decided to move into this space. Assuming VMW are working on a Type 1 in their labs a similar story. They also have nothing much on management and desktops. However I would be more inclined to think they would be more willing to move as a plug into vSphere and weakness in terms of bending over to MS at every whimper that there is overlap.

As for Virtual Computer and Moka 5 I think they are right in positioning themselves as management vendors that enhance a type 1. It makes little sense for them to become hypervisor experts if they can't make a dime, and OEM reach for them will be very hard and expensive. So I would only expect them to evolve the core hypervisor enough to help create a market for their management platform. With Virtual Computer I wonder if their management layer can be ported across hypervisors to give them broader appeal that just Xen? With Moka 5 I wonder why their management layer is not available on top of XC instead of their bare metal bs which is a joke? I assume Citrix *** blocking them in the hopes of doing their own slowly evolving, low quality, 3hitty featured version that will not meet the needs of most because they simply don't get enterprise requirements. I am sure Templeton will be talking more consumer crap at his next key note which he is not enabling for the enterprise with Client hypervisors because all Citrix knows how to do is HDX to more clients (AKA Receiver) providing a crappy windows display experience on devices that is really not that interesting beyond a short term tactical play.

So Tim, I doubt you will be happy in 2011 given the current vendor strategies. MS will be slow and may never move. VC, M5 etc are management strategies looking for Type 1 from elsewhere to take off. VMW will be also be slow but I assume will eventually move, and Citrix is too stupid to understand they need to provide a compelling management story quickly to gain market share and stay ahead. Why not buy VC of M5 Citrix/VMware? Clearly they get the space in management and are years ahead of anything you will pull off on your own. Let your smart people build the hypervisor right and please don't let MS win there, they will F the industry with it. Buy a management team who knows what they are doing, you all have plenty of cash.... then perhaps Tim can have a Merry Xmas 2012/2013.

Icelus wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 11:20 AM Link To This Comment

It's the waiting game for sure....

I think a lot of people know that making a type 1 client hypervisor is not as easy as making announcements for one.

Such an effort would only be suitable for enterprises rather than consumers due to the profits.

It took nearly 5 years for SHVD to be worth looking at so to think CHVD would just mature in the blink of an eye is just being impatient.

Here's something to think about:

Virtualization vendors went after business use cases for SHVD first then they started looking at CHVD to compliment it. Would there have been a difference in Desktop Virtualization acceptance if CHVD was the first focus then SHVD was the afterthought?

Going from Traditional Computing > CHVD > SHVD would be a much cleaner transition. Where everyone would do the first transition but specific use cases would permit the second transition.

A lot of similarities exist with Traditional PCs and CHVD, while SHVD bring a lot of "unnecessary" baggage.

Doug Lane wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 1:35 PM Link To This Comment

Don't be melancholy, Tim. There is a lot to be excited about. Compared to the engineering challenges we have already tackled in areas like performance and HCL, the last mile of usability items you have mentioned are chip shots.

I think the sheer fact that we were able to get within inches of being a day-to-day solution for one of the biggest Windows power users out there speaks to the fact the type-1 client hypervisor technology is ready for prime time when it comes to mainstream business use.

One of my resolutions for this year is to do a better job putting the spotlight on organizations that are using NxTop on a broad-scale basis to manage their corporate PCs. While client hypervisor-based systems are still the overwhelming minority compared to native PCs, there are IT folks running their shops of this technology today with great success. Stay tuned for some concrete examples from us.

I also think Icelus is spot on raising the question of how things would be different if vendors started with a client-hosted model. That is exactly how our company came to be. We looked at server-hosted VDI, liked it, but through it was starting in the wrong place.  Pretty much as simple as that.

In terms of appdetective's question about extending NxTop management to other hypervisors, it is a bit of a moot point right now since there are no other options that have significant traction. However, we did build our product to be extensible to other hypervisors. There is a cost to this, so we won't add another hypervisor on a whim. But if another client hypervisor opton catches fire, we plan to be proactive about it.

That said, we do see value in being hypervisor experts since it ensures us a seat at the table with the big guys and is also core to our management approach.  We don't have the luxury of a wait and see approach.

Doug Lane

Virtual Computer

appdetective wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 4:15 PM Link To This Comment

@Doug I really hope you guys keep pushing forward just like I do Moka5. If you r architecture is able to be ported across hypervisors then I think you are in good shape. Agree with remain experts to have a seat at the table, I just don't expect OEM providers to ship your hypervisor which is very hard to do for a small company and expensive. That said Splashtop has proved otherwise, so good luck and really looking forward to hearing about your customer stories.

Calvin Hsu wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 5:50 PM Link To This Comment

@Tim - While certainly there are more machines than we’ve officially qualified on the HCL that will work with XenClient, they do need to meet some minimum requirements such as Intel graphics, VTX/VTD. So it would help if you could spec out the machine you are trying to use and we can determine if it’s a bug or just simply not likely to work at this point. If you’d really like to do a full review, let me know and we’ll see what we can do about ensuring you get a machine that is fully compatible – it would be great for you to really see it in action before you dismiss it entirely. As Icelus says, these things take time to mature – however, “these things” are happening rapidly (we just put out the Service Pack 1, for example) and we are on the cusp of seeing issues regarding HCL, etc., diminish significantly.

You mentioned Brian’s article on the initial release of XenClient 1.0, but hopefully you also had a chance to see his notes from his in-depth interview of Peter Blum and Ian Pratt a couple weeks later (www.brianmadden.com/.../in-depth-q-amp-a-with-citrix-about-the-current-and-future-xenclient.aspx) – this discussed many of the questions and issues you raise in your post. For some reason this article doesn’t come up in the search and was a little hard to find, so I thought I’d bring it up again. There’s a ton of great stuff mentioned in there that has happened or is happening already.

Daniel Bolton wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 6:31 PM Link To This Comment

@Tim "The biggest hassle with installing a type-1 hypervisor on an existing laptop is the destructive install thing."....

If your referring to Nxtop here - and I think you are... You can use disk management in Windows to resize (shrink) your primary partition and then install NxTop... It will then automatically detect you have windows and install side by side - boot menu/loader included :)

I agree - the browser (and Quest/Citrix/other tools) at the hypervisor level is cool!

Like all things virtual - GFX are the main thing lagging for me... I think the next 12 months will be very interesting (hope anyway).

Dan.

Kimmo Jernstrom wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Mon, Jan 10 2011 11:10 PM Link To This Comment

Nice post Tim!

I' have about and around the same req.on what I'd like a CSH do for me.¨

It sure has been a pity and agony these couple of years with what we have. To be quite frank I've  postponed my type 1 dreams / delutions for the time being, this goes both for my "fun park" aswell as a, by any strech, viable solution in generali.

My "fun park" is tech-wise quite the same as 10 years back. Bare metal + type 2. By any measure that'll do. - However, I'm curious of the Hyper-V rig you're running.With the necessity of hibernation and proper sleep + power mgmnt on a laptop rig - why do you endure? Honestly, Is it because of the MVP thing? It makes absolutely no sense?

John Whaley wrote re: Disappointed! (the true sad story of client hypervisor meloncholy)
on Tue, Jan 11 2011 6:05 PM Link To This Comment

@appdetective We have looked at XenClient many times and have been disappointed every time.  We first looked at it over three years ago (when it was called XCI) and it looked promising but a long way off.  Then we looked at it again (when it was called Project Independence) and it was still nowhere close to the hype.  It was supposed to be released in 2009 but it wasn't.  In 2010 they released a beta which was a long way away from being something deployable.  Finally they released their 1.0 which is still something we cannot deploy - on the first day of trying it we ran into 4 critical problems that we couldn't in good conscience sell to customers as a solution.  XenClient will eventually get there (probably 2012) and when it does we will most likely support it (assuming customers are asking for it).  The concept is very attractive but the execution has been lacking.  And like Doug we don't have the luxury of waiting either.

And to Tim, honestly there is no good reason for you to be using a Type-1 on your laptop other than cool geek factor and wanting to play with technology.  You would be much better served by running a Type-2 on a Windows laptop - type-2 VMMs are much more mature, are easier to manage (snapshot/rollback/backup/clone/etc.), are easier to run on other existing machines, are more compatible and work with all your devices, etc.  There are only two good reasons to use a baremetal solution: to avoid having to manage the host, and if you have security concerns about the host, neither of which is true here.

-John

(Note: You must be logged in to post a comment.)

If you log in and nothing happens, delete your cookies from BrianMadden.com and try again. Sorry about that, but we had to make a one-time change to the cookie path when we migrated web servers.