Will bladed PCs affect Citrix?

In this article, I look at what bladed PCs are and how their use will affect Citrix.

Some major investment firms have recently published reports indicating that Citrix will have to compete with a fairly new technology concept called a “bladed PC.” In this article, I look at what bladed PCs are and how their use will affect Citrix.

What is a bladed PC?

All three big server vendors make server blades. (Dell , HP , and IBM ) Blade servers are basically like regular servers except that the form factor is a removable device that fits into a specialized chassis, and multiple blades share certain common components (CD-ROM, floppy drive, KVM, etc). The result is that bladed servers are much higher density than standard rack-mount servers (20 blades can fit in 3U of rack space), they consume less power, and they're simpler to manage.

Bladed servers are perfect for environments where you need a lot of similar servers, such as groups of web servers or MetaFrame server farms. In the latter case, companies using blades as MetaFrame servers and can fit probably 20-50 users per blade.

Recently, some analyst firms have been talking about how "bladed PCs" might be a threat to Citrix.

A bladed PC is the name given to a setup where bladed servers are used as end-user workstations. HP more-or-less invented this concept (at least from a marketing standpoint ), and it basically calls for a server blade in the datacenter running a desktop operating system with a thin-client device on the user's desktop. Conceptually this is similar to Citrix or Terminal Server, except that there's a 1-to-1 ration between active users and blades.

Citrix and Terminal Server technologies allow users to share single servers. With bladed PCs, the blades in the datacenter run Windows XP instead of Windows Server. The “remote desktop” capabilities of Windows XP are leveraged so that a user logs on to “remotely control” their blade.

The main advantage to bladed PCs is that companies can get the best of both worlds. They get the simplified management of thin client environments with the isolated user protection of traditional fat client environments.

What does this mean?

For the past ten years or so, Citrix has been laying out the advantages of the thin client device strategy. (They're more manageable, you can reuse old equipment, less breaking and maintenance, etc.) People have really latched onto this message, and Citrix does about $600M per year in sales to prove it.

Unfortunately, analysts see everything in black or white. They want to compare Citrix and bladed PCs in a “which one will win” type of scenario. (Or, more appropriately, they want to know how the bladed PC trend will affect Citrix's stock.)

The answer to this question is simple. Comparing bladed PCs and Citrix MetaFrame is like comparing apples and oranges. There are a lot of similarities, but at the end of the day, you have two different technologies for two different situations. Bladed PCs are not a Citrix-killer—they're an networked PC-killer. In fact, bladed-PC architectures actually complement Citrix server-based computing architectures.

Choosing Bladed PCs or Citrix

Citrix and other “traditional” Windows terminal server-based applications work best when a few applications need to be used by many users. As I've written in the past, every company in the world can use Citrix MetaFrame at some level. Some companies might be able to leverage it for 90% of their applications while others might only be able to use it for one or two applications. However, there are no companies out there today who can use Citrix MetaFrame for 100% of their applications in every scenario.

The reasons for this usually come down to engineering. It's usually not worth bringing an application into a MetaFrame environment if it's only going to be used by one or two users. In larger companies you might find 50 or 100 “fringe” applications that are only used by a few users and that will never be loaded onto a Citrix server. In the vast majority of Citrix's current installed environments, this “hybrid” application approach is used. The main corporate or department applications are deployed via Citrix, and the remaining niche applications are deployed traditionally.

Let's contrast this to bladed PCs. Bladed PCs provide the “simplicity” of the traditional PC model. One user = one computer. This is compelling for several reasons. First of all is the fact that because each user accesses a single blade exclusively, stupid user tricks only affect the user that screwed something up. No one else (since they're running on their own blades) is affected. Also, administrators don't have to worry about application testing, security, and compatibility with Terminal Server since each blade is running a single-user copy of Windows XP. Finally, printing, registry size, memory limits, and PDF-creation issues disappear typically found in Citrix and Terminal Server-based environments dissappear in bladed PC scenarios. (Of course there are third-party utilities that can manage the impact of all of these issues in Terminal Server and Citrix environments, but each of these products adds a layer of complexity while the PC architecture of bladed PCs have this built-in.)

Now that we've looked at the basics, let's look at some specific situations where a company would choose bladed PCs or Citrix server-based computing. Bladed PCs, Citrix/Terminal Server, and traditional desktops each have their own relative advantages. While there is some overlap, each solution offers some advantages that the other solutions do not.

Advantages of Bladed PC Environments

  • Users can access applications from any device over any connection
  • Broken client devices can be replaced quickly
  • No application integration issues
  • Stupid user tricks only affect one stupid user at a time

Advantages of Citrix or Terminal Server Server-Based Computing Environments

  • Applications can be managed on relatively few servers
  • Efficient hardware resource utilization
  • Users can access applications from any device over any connection
  • Broken client devices can be replaced quickly

Advantages of “Traditional” Fat-Client Environments

  • Users can use applications when they're not connected to the network
  • No application integration issues
  • Stupid user tricks only affect one stupid user at a time

Since each architecture has unique advantages that the others do not, companies need to look at this list and figure out what their best mix is. This is analogous to creating a blended investment portfolio. The best companies will blend all three of these technologies, and bladed PCs will “back fill” in some areas where Citrix didn't make sense. However, it's important to stress that the advent of bladed PCs won't prevent people from using Citrix where Citrix makes sense. If a company has an application that needs to be provided to a lot of users, the Citrix architecture is the way to go. Period. All of the benefits touted by Citrix still apply even when compared to bladed PCs.

At the end of the day, this decision boils down to one thing: money. In this case it's about which solution will be the easiest to manage (i.e. “cheapest”) for a specific portfolio of applications. The cheapest solution will vary depending on the number of users, the number of applications, the number of office locations, and the expected availability of the whole system.

This is really what all this (this whole “server-based computing” thing) is about. It's about creating the right toolset based on a mix of tools that can support an organization's application portfolio. Some people need a 700-piece mechanic's set while others only need a pocketknife. All-in-all, the invention of the screwdriver did not hurt sales of the hammer. Even though both screwdrivers and hammers are both used to build things, people still need the right tool for the right job.

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

This message was originally posted by Kimmo on April 6, 2004
Bladed PCs sucks...
1) You need another PC to access the bladed PC (1+1=2)
2) The bladed PC will be accessed thru network. GFX/Sound will suffer, aspecially with Remote/WAN connection
This message was originally posted by Anonymous on April 6, 2004
Is this why Citrix bought GotoMyPC ?
This message was originally posted by Brian Madden on April 8, 2004
Just a few clarifications. When bladed PCs are used, the desktop device is a thin client, not a PC, so it's not really 1+1=2. Also, I don't think anyone is suggesting using them over remote/WAN connections. Their use is limited, though, and that's the point I was trying to make in the article. They are cool, but only in the right situation.
This message was originally posted by Freddie on April 15, 2004
The blade PCs are made to be used solely in LAN and mostly as a second choice to fat-clients. Not thin-clinents...
This message was originally posted by Freddie on April 15, 2004
First of all, the blade PCs are made to be used solely in LAN and mostly as a second choice to real thin clients.
One thing that is really important in blade PC environments is that you add one more potential error source, simply because the solution is built on more hardware units.
Even if it can look like the blade PC are a threat to Citrix/TS they are more like an ad on solution. Now you can use the blade PC from a thin client to reach your specific apps. that just a few people in the company need and at the same time you can connect to your Citrix/TS farm via the blade PC and reach the base apps. In this way the blade PC becomes a cheap alternative to a fat-client. So in short, the blade PCs will be a larger threat to the PC world than the SBC world.
There for, the blade PCs will only be good for Citrix and TS, it will just introduce companies into the SBC world from an unorthodox way…good for us.
See ya…
This message was originally posted by Adolf Oberleitner on May 27, 2004
I see a great future for blade pc's in the soho market. a "full service provider" can offer manageged, easy to use, and "secure" solutions for the soho market.
This message was originally posted by Michael Burke on May 28, 2004
Anyone who's been around a while will remember a company named Cubix. They have been doing this since the 386 processor days. They are still in the blade server/PC market. Check it out at www.cubix.com.

Speaking of blade servers, anyone really interested in a hot product should check out Egenera (www.egenera.com). They are a blade architecture also, but what sets them apart is virtualization. They have "processor blades" that aggregate processing power in the chassis. Where HP, Dell and IBM's solutions are based on "one blade, one server", this solution is a dynamically allocated "PAN" or Processing Area Network. Processing resources can be dynamically allocated as needs change. Hot stuff...
This message was originally posted by an anonymous visitor on July 30, 2004
From their data sheet: Proprietary HP2 compression algorithms enable real-time remote visualization that is visually lossless
and has extremely low network bandwidth demands.
• Remote visualization: Complex 2D graphics are possible via remote graphics hardware acceleration.
Personally, we are going to use a blade environment to host Citrix. Why? Cost...it makes sense. Second, easier to manage. Will blades hurt Citrix? Not if people look at it from a pure cost persepective, they'll save money buying the blades.
Oh yeah, I totally agree here. Using blades for Citrix servers is awesome. The point I was trying to make in the article was focusing on the bladed PC concept, which does not involve Citrix at all.
I suppose this argument would get interesting if you looked at the feasbility of running VMware ESX server on a group of blades and potentially getting 10 or 20 individual desktop environments per blade. "Desktop" PCs could be easily provisioned, managed using active directory, centrally backed up and it would not require any more processing power or hardware than Citrix currently does. Of course you still have the issues with common PC peripherals and portability. However, consider a user being able to take their entire desktop environment with them as a VM running on a laptop while they are traveling. Any other ideas?
10-20 desktops vs 50-60 sessions. And why would they need a VM of their environment if they are using a laptop? The laptop should be their desktop. How do they get access to network resources when on the road? email? applications? As you said your still dealing with desktops and all the problems and issues that come with those. I personally don't see any advantage to doing it this way, I see more headaches and issues than even a TS environment produces. Maybe someone can convince me...

Why is ClearCube not mentioned in your article? You have left off the inventor and leader in blade PC technology. Your description above is inaccurate and misleading. ClearCube uses either a dumb user port or a thin client to access a PC, not a server. The benefit is similar to Citrix - but much more: centralized hardware, secure workstations, ease of management, ability to push out updates/apps/ etc remotely and TRUE lower cost of ownership while giving the enterprise 100% compatibility with all apps since they are running real apps, not published apps. Also, the enterprise is not paying 20% per year to keep their Citrix software updated, which most sites I've seen gloss over.
I think its been posted elsewere but you need to read in between the lines of the egenera marketing blurb. egenera is not as hot as it first seems. be careful...
Use the many hats approach:
Take off all the hats (practical, technical, user view, business view, religion etc) and simplify the problem by starting with just one hat. I start with technical and I usually get to the decision points quicker and all the noise becomes a little clearer. I see that all the options may not apply to me and could easily apply to someone else. So review the technoliges and understand minimally their differences and then walk through the problem with your situation as the data. Its the data that makes it the right or wrong answer, so it is assumed that when your done, you can then review your work wearing each of the other hats in turn.
Generically it looks like:
How many front-end functions (apps) can be served off a shared terminal server approach?   answer = X.
Of the remaining apps (Y), is the problem compatibility (running in 2003 etc)? or processing power / sandboxing? (a, or b, or both)
Would (for Y) running them on dedicated hardware fix sandboxing (probably), but would it be practical (dedicated/Win2003/Citrix/desktops? -eek)
What would be more practical, leave them on the front? (mixed mode, windows behind windows, multiple authentication, not achieving reduction in support, lose mobility as local apps are missing etc), or reproduce them on the back in a more sensible way? (I only need to run Y remember)
Could Y be hosted differently than X? does it need to be all or nothing? can I still get an integrated view to the front-end with x and y on the backend?
How do I achieve "my" OS choice on the back-end and not be dependant on my presentation vendor? (Tarrantela / Sun global desktop?)
Can this be segmented by classes of users and can I give each class a seperate solutions (assuming volume and practicality)?
Does the protocol (ICA, RGS, RDP) deliver the device and graphics richness I need? over what distance / load? can I support other devices?
Do other people need to be part of this discussion? when the user moves to a different desk, how do phones follow them?
There is a technology answer for almost any situation you come up with, unfortunately its not one single vendor / solution and needs re-packing inside your firm. Virtualization is in some ways a convergance point, but each of these vendors already has their own flavor of that, they need to play together. At least, if you are able to commit to a mostly back-end served environment, the client problems become easier - they just need to support software clients (ICA, RDP, HP RGS etc).
Sorry for the long post, but this is my Q4 2005 and most likely most of 2006.  =)
Steve: b.different@gmail.com
True...I was thinking that whilst reading the article.
Quite interesting. I am too late in this. But I think using VM ware for server is not a bad solution. Then u can use VM + server to serve many clients. These clients databases can be seperated and can be backed up. Laptop can call back server using thin-client solution like remote using a subcribed user name and password. U also can request to configure ur ram/storage/processor speed any time u like. It made all become a PC online service. Why not?
The only question I have left is that how efficient VMware is. Will it cause many overhead on server?
There are some environments where I pitched this concept when I was at HP, mainly the medical community.  Hospitals have some of the worst software imaginable.  They still use DOS, 16-bit applications!  Not always by their choice, the government sometimes dictates what apps they have to use.  Think about it, this is one of the worst applications to use in Citrix or Terminal Server.  However, it is perfect for a blade PC.  Any by not storing any data on the local drive, you meet some HIPPA requirements as well.  Unfortunately, development of these blades kept delaying the concept.  Not sure if anyone is even interested in this solution anymore.