HP Outlines Converged Infrastructure Architecture

Hewlett-Packard has fired back at Cisco in the increasingly contentious race to field an overarching data-center strategy, which will enable enterprises to rein in the complexity of sprawling networks and rampant virtualization.

Earlier this week, Cisco linked up with EMC and VMware in a partnership that tightly integrates storage alongside easy-to-manage virtualized data-center bundles called V-Blocks.
On Wednesday, HP unleashed its response. The computing powerhouse announced a converged infrastructure architecture that essentially creates a one-stop data-center shop. It packages HP's compute, storage, networking offerings into a highly virtualized bundle that can be centrally managed. The server components are HP's BladeSystems.

The converged infrastructure news was the highlight of a lengthy list of announcements bundled  into a press release headlined "HP Helps Organizations Thrive in Unpredictablity." HP spotlighted its BladeSystem Matrix as a proof-of-concept convergence platform, billing that package as an integrated shared services infrastructure for cloud computing. HP also rolled out consulting services to enable customers to field its new integrated products.

On the management front, HP unveiled Neoview Advantage, a new release of its enterprise data warehouse platform sporting real-time support.Wednesday's bullet points included the HP Data Center Smart Grid, which is a monitoring and  energy management adjunct to the infrastructure architecture lineup. It's intended to enable customers to dial down their cooling and electricity costs.]

While the dueling HP-Cisco intros are important in and of themselves, for CIOs what's of greater significance is what they say about where the metaphorical "head" of the industry is at. Namely, everyone is grappling with several competing, simultaneous and incredibly complicated challenges.

Leading the list is the rapidly ramping number of virtualized server resources, which many enterprises have to struggle to manage. In typical environments, it's often difficult to discover all one's resources, let alone perform live migrations, when compute power has to be reallocated on the fly. Indeed, it's becoming a truism that migrating a virtual server can often cost as much money as buying a new physical box.
 
Then there's the aforementioned network sprawl, which results from the explosion of server instances, both physical and virtual. One can add to this the impetus to boost network speeds to 10-GB Ethernet and beyond. Also in the mix is the rapidly vanishing network perimeter, where the traditional, well-defined boundaries have been exploded in the face of geographically dispersed, occasionally connected mobile workers. The requirement to support such remote clients adds security worries (aka authentication) as another unpleasant to-do item to the already strained checklist of admins and CIOs.

And I haven't even mentioned cloud computing. The potential capital-expenditure savings promised by moving hosted apps into a SaaS model is making chief financial officers salivate. Their CIOs, in turn, now have to contend with a mashed up environment where, as a first level, apps are bring shifted into the cloud. But additional complexity lies ahead, as organizations are being forced to re-envision their entire networks as a mix of public cloud resources leased from the likes of Amazon, combined with on-premise private clouds.

That's the multivarient backdrop against which HP, Cisco, IBM, Dell, Juniper, and others are now duking it out. Cisco has come up with what's perhaps the most enticing method--certainly, marketing-wise--of taking complexity and bundling it in under a buyer-friendly umbrella, via its Unified Computing System, announced back in March. UCS, which is both a product family and a strategy, bundles compute servers with network switching and management.

With its converged infrastructure architecture, HP is similarly trying to shield from complexity customers who are more concerned with running apps than managing the nuts and bolts of their compute resources.

"IT sprawl has businesses at the breaking point," Dave Donatelli, executive vice president of HP enterprise servers and networking said Wednesday on an HP Webcast intended to explain the converged architecture announcement. Donatelli noted that, by the end of the year, virtual servers will outnumber physical servers. "So now we have virtual sprawl," he said.

"It is clear that the traditional hierarchical network model is broken. It is clear that networks must be architected for virtualization from the start," said Paul Miller, HP's vice president of marketing. Indicating that HP is just as cognizant of the importance of storage as are Cisco and EMC, Miller noted that storage is another element of the equation, which hasn't kept up with the demands of virtualization. "We have to virtualize the entire storage stack," he said.
 
Beyond the compute and storage guts, management may well remain the trickiest element for competing vendors such as HP and Cisco to streamline, given that most of their customers operate highly heterogeneous environments. "Management must evolve to a shared-service model that is application-aware," said HP's Miller. He noted that shared-service environments span Unix, Linux, and Windows, and that dynamic failover of applications and disaster-recovery systems much work across all operating systems.

In summary, the Cisco and HP news of this week is likely only the beginning of a long series of data-center salvos. So for customers, navigating this environment is likely to get tougher, as they assess competing offerings.

Yet one wonders if perhaps the biggest problem is that the whole concept of data-center sprawl, and the management thereof, remains obtuse to the initiated. As in, how the heck do CIOs explain this stuff to their CEO?

Follow me on Twitter.