HP's aggressive "Blade Everything" marketing campaign has motivated some customers to ask whether or not we agree with HP that blades are the only choice for today's enterprise. In a nutshell, we don't agree. We think the strategy is impractical for customers looking to reduce IT cost and complexity and aiming to take maintenance and operations to 30% of the IT budget and innovation to 70% (see our last "Dear Mr. Hurd" post). CRN's Ed Moltzen weighed in on the topic and Chinese reporter and blogger Sun Yongjie also provided perspective as well.
With good reason, customers are excited about the blade server proposition (and I'll tip my hat to HP for doing a nice job spreading the blade gospel). Why all the excitement? Appropriately deployed, blades are easier to manage and service; they're more energy efficient and offer greater density. What's not to love?
But, we believe a "Blade Everything" philosophy is not in the best interest of our customers. In fact, we believe that applying that philosophy could actually increase IT complexity. That is completely counterintuitive to Mr. Hurd's and our shared goals of simplifying IT and our 30/70 goal. We'll continue recommending blade servers to customers when it makes sense based on their real world problems. In essence, our mantra is "Blade When Appropriate," blades should be an option… not a mandate, and I think 99 percent of our customers and the analysts would agree that blade servers aren't for everyone. Even blade-bullish IBM's Robert Ames agreed in a recent Government Computer News article.
So what else are our customers telling us about their needs and blade servers? I can boil it down to four key points:
1) Customers want simplicity: One of the top issues I hear every day from customers is they want flexible systems that will grow with their organizations. The last thing they want is to be locked in to one vendor's offering. Blades by nature are a proprietary infrastructure due to lack of interoperability between vendors' chassis (unlike optimized rack mount servers). When you start adding multi-core architectures to the mix (i.e. 8, 16 and 32 core offerings HP and other vendors are driving towards), blades simply become a veiled proprietary architecture, the "next generation mainframes" if you will, designed for little more than single-vendor lock-in. Couple that with some of our competitors' proprietary back-end software offerings such as HP's OpenView and Dynamic Thermal Management tools and customer quickly see that in some cases this lack of flexibility can increase IT complexity. John Enck, a research VP at Gartner, captured the lock-in problem nicely in an internetnews.com story.
"We don't see blades becoming world dominant," said Enck. "They have their advantages, chiefly density. But the IT guys know they have a choice of adding rack servers from Dell, HP, IBM and others. If you go with a blade system from HP or IBM, you're pretty much locked into their environment."
In addition to vendor "lock-in," the "Blade Everything" mantra should also include "Upgrade Everything," including customers' power and cooling capabilities and the budgets needed to implement them. Computerworld put together a piece on blade best practices earlier this year that touches on many of the issues customers should consider. While the article did a nice job capturing many blade issues, a comment left by Jim Armstrong really drove the point home.
Wholeheartedly adopting blade servers would require some customers to build entirely new data centers with greater and different power sources such as 220V to each rack which isn't standard in many data centers, and cooling capabilities. That makes no sense. Customers want to build their business not new data centers. They rely on us to develop products and solutions designed to meet their goals and objectives. Many times, the solution doesn't include blade servers, which is the why we'll continue to invest in innovation in both blades and mainstream optimized rack products.
When a blade server does make sense for an organization, we're happy to supply a product that's easier to deploy, has greater performance/watt than HP's latest c-Class line-up and also includes back-end software that's open-ended and built on industry standards to fully integrate into the tools customers already use for management (such as Altiris, Microsoft MOM/SMS and Novell Management software).
2) Ease of deployment: Our customers' aspirations to achieve 30/70 says they want to spend more time on innovation and less on management and maintenance, which would include fumbling with deployment. Customers want fast deployment right out of the box. To test how we were doing, we recently commissioned a comparative report (Dell's existing blade design vs. HP's new c-Class blades) The third-party analyst that conducted the study ordered a chassis full of blades off of each of Dell.com and HP.com just like a customer would. To summarize, the report found the typical implementation time for 10 Dell 1955 blades (a full configuration) was roughly 1 hour and 5 minutes versus 3 hours and 34 minutes for a full configuration (16 blades) of HP c-Class blades. We also saved a few trees by only shipping Dell blades in 2 boxes, instead of the 78 boxes shipped by HP.
Good server administrators are difficult to find and demand high compensation. The math is easy: HP blades take 75% longer for an administrator to deploy. Imagine the impact to a typical 300-unit blade deployment. That's valuable IT manpower that could be focused on innovation…not maintenance.
3) The marriage of energy efficiency and performance: We have personally seen the effect rising energy costs can have on data centers and consistently hear that customers want to improve efficiency without compromising performance. Our blade R&D team has been hard at work finding ways to increase performance with less energy, and I was happy to see that a comparative test showed that we've made great progress. It showed that our current offering — PowerEdge 1955 Blade server — drew less power (which of course means less heat) and performed better per blade than the new HP BladeSystem c-Class The Dell 1955 system achieved better performance/watt at every configuration tested regardless of blade count. I wonder what HP's "Blade Wonk" thinks of those numbers.
You'll see our blade advantage growing later this year even more. Here's a recent post where you can get a sneak peek on what's coming.
4) Data center vs. Platform perspective: The last point we hear from customer may be the most important. They are focused on long-term goals and projects designed to optimize their data centers and not just one specific server or application. HP's "Blade Everything" campaign puts way too much focus on the platform instead of real customer pain points like IT complexity, power and cooling.
Instead of subscribing to the "Blade (and Upgrade) Everything" mantra, we believe it's in customers' best interests to evaluate what makes the most sense for their businesses. Sure, shiny new blades can provide an enormous amount of value in total cost of ownership, but for some customers the decreased flexibility and increase in infrastructure costs required to power and cool this density may completely negate TCO benefits. "Blade When Appropriate" proposes that customers look at a variety of options including blade and rack-mounted servers when evaluating solutions.
I believe the "Blade Everything" approach will actually increase IT complexity and make it more difficult to reach customers'30/70 goals. This clashes with where CIOs and their teams want to focus today.
As always, I'd love to hear any additional insights or ideas you might have about blade servers and the 30/70 innovation to maintenance imperative. Please feel free to comment on this post, head over to IdeaStorm or blog about it on your own and tag it "Dell30/70" so we can find you.