Win the desktop, and you will win the server

Or, "Why Red Hat is pursuing the wrong business strategy"

Red Hat has recently announced that they have "No plans for a traditional consumer desktop". Let me explain why I think Red Hat needs to change their business strategy.

First, a short history lesson. Before the arrival of Windows NT Server, Novell Netware claimed 90% of the market for PC based servers. However, Netware made a near fatal mistake when they did not provide a GUI interface soon enough. This comes from the same Wikipedia page linked above:
While the design of NetWare 3.x and later involved a DOS partition to load NetWare server files, this feature became a liability as new users preferred the Windows graphical interface to learning DOS commands necessary to build and control a NetWare server.
So server administrators became familiar with Windows 95 on their desktop, and they naturally preferred Windows NT 4.0 which included the same interface.

Challenged by Ubuntu

Red Hat is in a similar position to what Novell faced, in that Red Hat is facing a time when server administrators will choose to run their desktop operating system on their servers. Specifically, I believe that Ubuntu will soon become the de facto Linux desktop. This means that server administrators will become familiar with Ubuntu and develop a trust for the brand. Eventually, they may choose to migrate to an Ubuntu standard on servers and desktops.

Most people agree that the real money is in server operating systems. If Red Hat wants to keep capturing that server money, they must provide a supported, free desktop operating system as part of a loss leader strategy.

Fedora is great, but it doesn't solve this problem

Don't misunderstand me, I know that Fedora is an excellent piece of software, but it has two fundamental problems. First, most average computer users do not know that Fedora is sponsored by Red Hat. This means that the Red Hat brand is not directly benefiting from the popularity and success of Fedora.

Secondly, Red Hat does not provide any support for Fedora. This means that many business cannot seriously consider running it on their desktops. How will these business get support if a problem comes up? How will they know that their applications are certified to run on Fedora? What if they want long-term support for older versions, without having to upgrade all the time? All of these questions are being answered by the Ubuntu ecosystem.

Red Hat, let me give you a hint. (If you want more hints, I am always available for consulting). Here it is: Change the name of Fedora to "Red Hat Enterprise Desktop" and begin to sell support for it. If you are lucky, it may not be too late to capture a large percentage of the desktop operating system market.

Remove the need for CentOS

Red Hat, I will even give you one more hint for free. Why do let CentOS steal your thunder? You have already published 99.999% of CentOS (everything except the branding). You graciously publish the source code to RHEL to abide by the GPL, but then you let another brand take credit for your work. How can you fix this? Easy! Simply provide a free version of Red Hat Server that is compiled and ready to be installed. Now your users will see even more of RedHat. RedHat on their desktop, RedHat on their servers, and they can buy support for all of it if they so desire.

Comments

  1. That makes a lot of sense. Although, why should we help Red Hat rather than letting Ubuntu win? :P

    All kidding aside though, Ubuntu does lack some of the nice GUI utilities for server configuration that Red Hat has already, although this seems to be changing somewhat. I'm also pretty interested to see where things go with ebox, which could be a nice in-between for those of us who like the convenience of a GUI, but still think running a full desktop is a bit silly.

    ReplyDelete
  2. Not to mention that some people want a distribution without support or obligations. Free edition would decrease the barrier for those people to use RedHat. It's not like if they were buying the commercial product anyways.

    To talk of the prices and the current editions.. They are awful. Artificial limitations (CPU amounts, memory, etc) and the prices are 2-5x more than what they should all aboard.

    ReplyDelete
  3. I'd like to *extremely* disagree with you on your point about CentOS on your recent post. CentOS is in no way related to RedHat, they make an almost exact replica of RHEL. RH doesn't mind that CentOS is there because when you realize that those who use CentOS will need support. When this happens, who are they going to go to? Novell? Canonical? No way! Its RH because they're already using basically the same thing.

    I also strongly disagree that winning the desktop will help win the server market. The server market might be low-hanging fruit, but RH also sells RHED already and its not taking off.

    This may sound absurd, but the only reason Canonical and Ubuntu are getting the Linux desktop market is because they got their first and they marketed it better than anyone. In addition, the release of Windows Vista helped.

    I will say though that Ubuntu is a nice distro and has come a long way. Problem is, they don't have any way to make serious money from the desktop market until their server strategy is in place.

    Cheers,

    Clint

    ReplyDelete
  4. Why would we want redhat or ubuntu to win when debian could win.

    Greater stability, more applications within its repository. I'm sure I could go on, but a stronger debian for me is the way forward.

    ReplyDelete
  5. I want to reiterate what Clint said.

    1/ Go reread RH's press release. They do have a desktop but they will only sell to business. Nothing wrong with that strategy. That being the case, it obliterates your argument of desktop --> server drift theory.

    2/ Your desktop wins the server example of Microsoft over taking Novell is in error. I need to point out that it was gateway services into Novell LAN's that won the day. WFWG was the dominant client at the time. Win95 came later.

    3/ Never assume that eyeballs equals consumer success. The worst thing a company could do would to be too successful at its marketing that the company drowns in service calls from Ma Frickets. RH is making a measured scalability decison that makes sense.

    4/ Go out to the VPS hosting market. RH/Centos are dominant. Ubuntu comes in a distant 5th. It will probably stay that way too. SOA is becoming a overriding fact of life in companies. SOA decouples the desktop from the server/services being provided. Fact its more important to have the right browser in a SOA world than necessarily the right desktop.

    ReplyDelete
  6. Redhat Enterprise Desktop? That's a great idea! I wonder why they haven't thought of it since Redhat 9..

    ReplyDelete
  7. My bad for missing the existing Red Hat Enterprise Desktop. It must have some major marketing problems if I had not heard of it before. The next question is, "Where is the CentOS-version for the desktop?"

    Maybe I should have changed my title to...
    "Give away the desktop, win the server".

    We shall see if RedHat can continue to thrive on the server if a majority of users are running Ubuntu. Only time will tell!

    ReplyDelete
  8. An even better lesson is Unix on the desktop. Back in the olden days there were many companies that made Unix desktops (aka workstations). As Windows (in the 3.x days) became popular, all the Unix companies looked down on it and retreated to doing servers. Windows migrated from being desktop only to also taking on the server world. It was only Linux that halted that advance.

    Redhat also suffers from not having a free enterprise desktop. Quite simply software developers typically develop and port to what they use first. Then they check other operating systems, distributions, processors etc.

    All the work I have done in the last few years has worked best on Ubuntu because that was my desktop (and servers). One customer asked for RHEL support and it was a huge pain just trying to find the OS so I could compile packages etc on it. CentOS is ok, but not the same thing. There is no way I could give the customer something compiled on CentOS that had never run on a real RHEL. When it came time to make VMware appliances, Ubuntu ended up being the underlying distro. I didn't have to worry about activating it, linking to support contracts or the availability of updates. It just worked.

    I used to work for a Unix vendor who decided their strategy was to be a "port to" platform rather than a "develop on" platform. They stopped being relevant about a decade ago as a result.

    ReplyDelete
  9. > Secondly, Red Hat does not provide any
    > support for Fedora. This means that
    > many business cannot seriously consider
    > running it on their desktops. How will
    > these business get support if a problem
    > comes up? How will they know that their
    > applications are certified to run on
    > Fedora?

    I couldn't disagree more. I mean, Red Hat provides Red Hat Enterprise Desktop (which is really stable and fully supported). So why should they call Fedora Red Hat Enterprise Desktop when they already have a stable enterprise desktop?!

    Also, I personally, as a businnes, would want a distribution which is developed by a company and which is fully supported and not something like Ubuntu or Fedora in which it depends on individuals to do security fixes (see universe) or which is not rock stable.

    Lastly, why do you think Red Hat should trash their working businnes strategy in favour of Canonical's not yet working strategy?

    ReplyDelete
  10. > The next question is, "Where is the CentOS-version for the desktop?"

    Just download the CentOS Dvd, put it in and select "Graphical Interface / GNOME". There you go, you have the shiny Centos desktop.. Btw., I am using it on some machines with great success because it is very polished and stable IMO.

    ReplyDelete
  11. Why Ubuntu users are so arrogant and ignorant? ... i mean, without RedHat contributions to Linux kernel and a myriad of projects Ubuntu would be ZERO, without Debian itself Ubuntu would be ZERO, without an absurd 'Marketing-Hype' Ubuntu would be ZERO (no real innovations or technical merits on Ubuntu's side -except ONE event daemon, upstart- OR contributions to the kernel, and that's a fact), etc, etc.

    So, stop being so arrogant, mainly because 95% of Ubuntu users were Windows users only a couple of years ago and they think that now can give some advices to one of the most important companies on Linux history ... damn. I'm really sick about all of this, :/.

    ReplyDelete
  12. RHEL 5 IS A BROKEN MESS

    I recently switched from RHEL 5 to Kubuntu, and I'm impressed.

    The Adept Manager actually works with zero dependency problems (try that Redhat).

    I filed yum and "Redhat Network" bug reports for years, and they were ignored by Redhat.

    Redhat is obsolete in my opinion, as they screwed their users years ago. I know others have switched also.

    ReplyDelete
  13. no offense. But even if netware had 90% of pc servers... windows still isn't top dog in the server market (although they have been making inroads). Back then x86 really wasn't used for servers.

    ReplyDelete
  14. Please read and understand the GNU + GPL license. Learn and study the relationship between Red Hat, Fedora and CentOS. Erase the stupid ideology of trying to dominate Desktop and Server like Microsoft, which still tries but continues to shoot their foot. Search and seek the history of Linux and how it has become to be what it is today. Think and re-evaluate your thoughts. Then re-write the article + all your posts because none of this make any sense. IMO - Red Hat should continue to gain the same market it has been working on for years, servers. Leave CentOS for us uber geeks and Ubuntu continue their strong Desktop road map, + servers if needed, because a lot of people think that they need to wear a Nike head bandanna to match their Nike shoes.

    ReplyDelete
  15. I think the article misses a crucial point.

    I question the premisse that server administrators necessarily want a desktop environment. Let me explain.

    I'm pretty sure they want to have *control* (from installation to monitoring to maintenance to updates to upgrades), as easy and as efficient as possible.

    A good GUI could be very helpful in providing that, presenting the data, and allowing it to be manipulated. However ... unattended installation using scripts and maintenance using scripts is very efficient too, especially when you think of a hundred or so servers.

    When you think of putting a GUI on top of that, you are talking about an *application*, not a facility in an OS. And that application might very well run under an X-windows desktop environment. But that doesn't mean that companies like Red Hat would stand to gain anything by incorporating a GUI into their offerings. So ... I disagree with your assessment that success in the server market depends on whether the server OS has a GUI attached.

    ReplyDelete
  16. Wow, it seems that most of the people commenting here misunderstood what you wrote (e.g. Clint, and the last comment before this one). I don't know why, I thought it was clear.

    ReplyDelete
  17. The author may be right for small deployments, but not for the big deployments. For firms like Novell, RedHat, IBM, ... there is little to be gained with small deployments. There main income comes from consultancy and maintenance to big clients. Short, from the analysis to the actual deployment of large scale projects.
    Ubuntu may prove to be succesfull for small deployments, but if they succeed on large scale projects, it will not be via direct contract sales, but via big names like IBM who can and will be responsable and accountable for large scaled projects.
    Would you trust a project to Canonical which is worth 1/10th of the total capitalization of Canonical?
    Neither will big clients. At best is the input from admins a parameter in the decision taking, but at the end, the decision is taken by others who are accountable. And since the risk is double of the contract, they want a thrustworthy (financially) party to deal with, not necessarely the most popular one.
    Companies like Novell, RedHat, IBM, Sun, ... should put their energy in the segment which give them the biggest return. They own that to their shareholders.
    And what is deployed will not be decided on what people run at home or is the most popular, otherwise MS would own that segment to.

    ReplyDelete
  18. There have been some great comments to this post, so I want to thank everyone for participating. I didn't intend for this to be a controversial post, but rather a discussion of strategies.

    It is true that companies who are managing thousands of servers will likely not be running a GUI interface on those servers. However, the GUI interface is useful for smaller server deployments, which account for a majority of server installations.

    One question I have is whether small businesses will even want to host their own server in the future. Instead, they may choose to use a SAAS hosted service. If SAAS is the future, then most companies will not be managing their own servers. Instead, the SAAS vendors will be managing a large deployment of headless servers, and they won't care about a GUI.

    With that in mind, it is possible that familiarity with Ubuntu desktop will NOT lead to a large adoption of Ubuntu server.

    I am a supporter of Red Hat, and I hope they succeed in business. I don't really care what brand wins, I just want open source to thrive and produce the best software possible.

    ReplyDelete
  19. Seriously Tristan, you DO NOT GET IT...

    First your Hyperic is so awesome use it and its java agent that will make the oom-killer go nuts on your box and Servers should be gui-only post... Now this?

    If the internet required a license to blog yours should be revoked. Time and time again you show your extreme lack of understanding in open source or the Linux community as a whole.

    Please stop doing the community a disfavor.

    ReplyDelete
  20. Good Job! :)

    ReplyDelete

Post a Comment

Popular posts from this blog

Using the Cisco console in Linux

Linux NIC teaming recommendations

What it takes to make Ubuntu ready for use