R.I.P. EBS: On-Premise Servers Not So Essential After All

Microsoft this past week announced that it would discontinue its Windows Essential Business Server (EBS) effective June 30, 2010, killing off the product after only a few years in the market. Honestly, I'm not surprised: EBS was a big, hairy, complicated product that targeted only a very small audience and, most likely, simply confused customers.

Not that it wasn't well intentioned. I first heard about Microsoft's midmarket server offering, codenamed Centro, way back in 2005, when it was still unbranded. At the time, Microsoft had two mainstream server lines, Small Business Server, which targeted businesses with 25 or fewer PCs, and Windows Server, which targeted everything else. SBS, of course, simplified the deployment and management of Microsoft's server offerings; not just Windows Server itself but also Exchange, SQL Server, and its network security offering. The idea behind EBS was obvious: Ape the success of SBS, but for the midmarket.

The midmarket, alas, was a tough sell. This market consists of businesses with 25-500 PCs, and in 2005, there were supposedly 1.4 million of them. Unfortunately, the complex part of SBS--the sheer number of bundled products--was amplified in EBS, which would utilize three or four physical servers, depending on the product edition, with the various products spread out between them. These would include a management server, an edge server, a messaging server and, optionally, a database/LOB server.


Because EBS was based on Longhorn Server (which became Windows Server 2008), it faced a lengthy time to market. The first beta shipped in 2006 and featured a torturous and integrated Setup routine spread across multiple DVDs, requiring 8 hours to progress through about 20 steps. (This was, however, an improvement over installing the servers separately, which could take 80 hours and 140 steps, according to Microsoft.)

As part of what later became a Window Essential Server Solutions family of products that included Windows Home Server and SBS, EBS provided an extensible, central management console with environment monitoring capabilities. (That work, I think, will live on in future products.)

And as with many newer Microsoft server products, EBS was x64-only, which makes plenty of sense. But it made testing EBS difficult, especially early on: You needed at least three x64-based servers or a virtual environment that could handle 64-bit guest OSes, a rarity just a few years ago. Speaking of which, Microsoft actually licensed EBS for use in virtualized environments. When I asked about this curious move, I was told that three server blades tended to be the break-even point on blade hardware, so Essential Business Server was coincidentally perfect for this use. Pricing and licensing, of course, was complex.

Ultimately, the setup and management integration in EBS wasn't enough to attract a lot of customers. I think the reasons for this are somewhat obvious, and define the midmarket itself. While there are many, many small businesses and a small but lucrative enterprise market, the midmarket is sort of nebulous. This tweener market is an awkward meshing of the cash-strapped small business market with the needs of the enterprise. And it's small, too.

Microsoft says that the market changed in the four years since it announced EBS. "Midsize businesses are rapidly turning to technologies such as management, virtualization and cloud computing as a means to cut costs, improve efficiency, and increase competitiveness," the company noted in a blog posting. I have to wonder what that means for SBS as well: If ever there was a market that doesn't need to be hosting its own email infrastructure in-house, for example, it's small business.

(And fear not for those that did buy into EBS: Microsoft is doing the right thing by EBS customers by allowing them to get the individual components of the product for free, for a limited time. Check out the Microsoft web site for details.)

Looked at from a broad perspective, I think what this boils down to is complexity and cost. And with more and more businesses eyeing offloading infrastructure services like email, PIM, communications, management, and file sharing, Microsoft and competing cloud computing solutions just make more sense.

I expect to see some dramatic changes in the next SBS version that reflect these needs as well. The age of the on-premise server isn't over, not yet. But this may one day be remembered as a major and notable milestone in our march to the cloud computing future. In retrospect, it's actually kind of obvious.

An edited version of this article appeared in the March 9, 2010 issue of Windows IT Pro UPDATE. --Paul