Goodbye Hardware, Hello Data Center Infrastructure

Arthur Cole

All right, so we all know that the data center is about to become more flexible, efficient, responsive and all around better than it is now through the convergence of virtualization, advanced networking, cloud computing and a host of other technologies.

The central question, then, is not "Where are we going?" but "How do we get there?" Not only will individual enterprises have to select the right technologies to suit their particular needs, but they will most likely have to sign on to one of the major platforms coming down the pike. While HP, IBM, Cisco, Oracle/Sun, Microsoft and all the rest tout their commitments to interoperability and open platforms, the fact is that the most direct route from here to there is through one turnkey vendor.

A leading candidate for many of you will be IBM. The company recently expanded its relationship with Brocade to package Brocade switches as part of part of the IBM's overall Dynamic Infrastructure platform. The deal specifically brings in the SAN director and DCX switches, allowing customers to choose between them and the Cisco networking equipment that is already available. Unusual, isn't it, that IBM turned to Brocade just a few weeks after Cisco launched itself into the server market with its own Unified Computing System?

Moving along a similar vein is HP, naturally, which came out with its own Adaptive Infrastructure last year and has issued a steady stream of upgrades to the platform ever since. The company recently introduced the BladeSystem Matrix server designed for rapid deployment of highly virtualized environments. And earlier this week, it announced that it was integrating its Insight Control management software with Microsoft's System Center suite as a means to provide broader control over those environments.

Cisco, for its part, is coming at the new data center from its traditional networking perspective, rather than the server/storage-centric view of the traditional hardware manufacturers. Its UCS platform relies on the company's 10 GbE prowess to form the fabric of future LAN, SAN and other networking architectures, bolstered by the new B-Series blade server that the company says is optimized for virtualization. As a networking specialist, Cisco supposedly has the upper hand when it comes to moving large amounts of data from place to place, a crucial component for those interested in advanced operations like automated load balancing and disaster recovery.

To get to the source of all this networking activity, however, one need look no further than Intel. The company has spurred a renaissance of sorts with its new Xeon 5500 processor and is now bolstering the platform with the 82599 10 Gbps network controller that the company is pitching for high-speed NICs, mezzanine cards and LAN-on-Motherboard systems. The company's aim is to see organizations begin to consolidate 1 GbE ports into 10 GbE ports, doing for networking what virtualization has done for the server farm.

Efficient networking has always been one of the trickier aspects of data center architecture, and if one thing can be said about these new offerings, it is that they provide a more streamlined, efficient way of doing things. But to gain that kind of simplicity, you'll first have to undergo a fairly complicated reorganization.

Add Comment      Leave a comment on this blog post
Apr 28, 2009 6:41 AM Ken Oestreich Ken Oestreich  says:

Arthur -

The real story here is that "orchestrating" the infrastructure (I/O, networking, etc.) is the perfect complement to virtualization.  Abstracting the rest of the data center will provide the agility we've all been waiting for.  HP, IBM and now Cisco have all acquiesced this -- building on what Egenera has been doing for ~8 years now.

Apr 28, 2009 9:42 AM Don Nalezyty Don Nalezyty  says:

I definitely agree the question of the day is, "How do we get there?"

Ken's comment above about orchestrating the infrastructure being a complement to virtualization is very true and that's the piece that's been missing.  We've been asking this question internally for a few years, but the technology just hasn't been there, so we've been forced to patch together solutions that haven't had the level of integration and flexibility we really need to run an entire enterprise data center that is highly heterogeneous with a vast variety of hardware and software. 

I think there are now enough offerings out there around data center infrastructure automation that we can not only start to answer that question in a meaningful way for the first time, but might have a complete answer.   That answer will fundamentally changes how the majority of data centers are run and supported.

That said I think there are still gaps in many of the existing solutions and that there's still a significant need for real innovation on the part of IT practitioners to bridge those gaps.  The rate of change in solutions in this space is definitely picking up and I have high hopes, that by this time next year we'll have moved on to the next big problem...


Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.