Server vendors’ days are numbered


opinion This week I had a conversation with an Australian chief information officer which I considered both profoundly interesting — but also extremely disturbing.

I was speaking to Komatsu general manager of IT Ian Harvison about his company’s recent decision to stop operating its own datacentre and server environment and shift its servers into an infrastructure as a service offering provided by Telstra. At one point, I asked Harvison about the underlying server platform which Telstra was operating.

His answer was nothing short of revolutionary.

“We’re not interested in the technology at all … from our perspective, we don’t care what the hardware is at all,” he said. And the executive went even further. “We’ve negotiated quite robust service level agreements with Telstra,” he said. “As long as Telstra delivers the SLAs, who they partner with, and what they’re running on, is a Telstra-driven initiative.”

Shocked, I laughed and pointed out to Harvison that server vendors — companies like Dell, Sun Microsystems, HP and IBM — would be appalled at the statement that he had just made. After all, these companies have been selling servers in the dozens and even hundreds to companies like Komatsu for decades now.

But Harvison didn’t blink.

“At the end of the day, their relationship needs to be more with the cloud partners, as opposed to the end customer,” he said.

Now, to some degree Harvison and his team will maintain an interest in what kind of hardware platform which Telstra is providing. For example, the executive mentioned that he had been interested in what storage systems Telstra was running.

In addition, Komatsu has negotiated what it calls ‘technology refresh windows’ with the telco, where it will meet with Telstra and discuss the incoming wave of new technologies, with the aim of ensuring there is a common understanding about the future development of Telstra’s platform and the governance of it.

Harvison said the last thing his company would want is to get stuck on old technology being used by its supplier. And the converse also applies — his company wouldn’t want things to advance so fast that its applications wouldn’t be supported by underlying infrastructure — such as Telstra’s virtualisation platform, for example.

However, the fact remains, that Komatsu is now a concrete example of a major Australian end user which does not give a flying $%^& about what server infrastructure is powering its more than a hundred virtual machines. This is a CIO of a major Australian organisation who simply does not buy servers. You’re wasting your time calling him, if you’re a server vendor.

And that fact must give any vendor who sells into the datacentre pause.

The reality is that in future, there are going to be much fewer discrete datacentres operated by major Australian organisations. Things are consolidating fast. Harvison’s comments represent a sharp warning to enterprise IT vendors: You had better start considering what your strategy is for selling your solutions to large service providers.

Right now, the list of service providers selling the sort of infrastructure as a service or private cloud solution which Komatsu is buying is small — some of the names on the list include Telstra, Optus, Fujitsu, CSC and so on.

However, the list is growing fast — in direct proportion to the list of companies like Komatsu who are dumping datacentres altogether.

For many Australian organisations over the next few years, the question may no longer be how many servers they need to buy. The question may instead be: Will they buy any servers at all?

Image credit: John Seb Barber, Creative Commons


  1. I wouldn’t say they are numbered. Data centres are going to require more servers, with more efficient power usage, more scalable storage, connectivity and such,

    not including your example there, look back 5-10 years, where an SMB would need 4-6 servers, Firewalls, storage, domain controllers, exchange, etc, and now can comfortably run on one Quad Core, 64bit system with 12GB ram, while still using less power, less $$, and equal or better operations.

    server vendors are going to have to refine their sales, and be even more competitive to stay ahead, and be the choice for Hosting co’s who need the best

    • True, servers will always be required — and we are still building larger and more complex datacentres. I guess what I was trying to say here was that server vendors are going to need to refine their models — things are changing dramatically from the way they were over the past few years.

      Another example of this would be the way that companies like Cisco are partnering with providers of other technologies like VMware and EMC. Servers are no longer servers — they’re more integrated these days with lots of other technologies.

  2. Wow, overly dramatic headline.

    Unless service providers suddenly start building their own servers than the vendors are not going away. They will have to adapt their products to address the different needs, but how is that different from the continually evolving landscape we’ve had in the past?

    • True, the headline wasn’t quite what I was trying to say here — it seemed like a good idea at the time. As Ben says above, I was trying to get to the idea that the way server vendors have done things for many years is dramatically changing.

      I think another factor is, as I mentioned above, the increasingly integrated nature of the datacentre. Stand-alone servers are increasingly going to become a rarity rather than the norm.

  3. The bottom line is many applications are ideally suitable for virtualisation and/or “cloudisation”, but many are not.

    Customers I deal with – (including large ones) – are mostly interested, but there are many pitfalls and twists and turns in the cloud market as it continues to mature.

    While it is certainly true that the major hardware vendors have to change their focus – (they can’t ignore the cloud) – the non-cloud space still has a lot of legs.

  4. Actually Renai I think that you’re being overly defensive as I agree with your overall sentiment and the trend you’re illustrating by reference to Komatsu.

    While the server market is booming it is booming now because of private clouds, when it comes to public cloud providers they have (1) far great efficiencies (better then the 61% unused capacity in average large private server farms, and (2) many don’t buy branded boxes and won’t pay for brands. I wrote this “Storm clouds on the horizon for vendors”

    Therefore these is a serious threat to hardware vendors for which they will have to readjust.

    We know that for Microsoft they see that the real benefit of its Azure cloud service is more than elasticity and scaling and economies of scale; it’s that you can have Windows servers without having to do the work of running Windows Server yourself. That’s the whole direction, and it’s moving up in functionality.

    I think that Komatsu’s playing a bit ahead of the pack, and well done.

    Walter Adamson @g2m

Comments are closed.