On the horizon is a profound change to the current data center fabric, possibly the most significant IT shift since the widespread adoption of the LAN itself: unification of the IP and storage networks.

We know what you're thinking: Moving to a new data center architecture sounds like the kind of grand plan to shelve for happier economic times. But given the potential long-term cost and operational flexibility benefits, CIOs and their top IT architects must plan now for the impact of a unified infrastructure. The move entails more than just running storage traffic using Fibre Channel over Ethernet. It's about creating a single, unified network fabric built on 10 Gigabit Ethernet that carries any traffic in the data center, from data to voice to storage to high-performance clustering.

The problem with today's data centers is that they're "islands of virtualization," Cisco CTO Padmasree Warrior writes in a blog post, with virtualized servers, storage, and networks all managed separately. In her much-discussed post earlier this year, Warrior laid out Cisco's vision for integrating those islands. Cisco is expected to provide details on its data center unification strategy, including new products, this week .

Is Cisco right? It's mapped out an inevitable path, since the opportunity to harness virtualized IT assets for agility and automation is too great to pass up. Which vendors are at the center of that architecture, however, is a question that has a long way to play out in the market.

A unified data center infrastructure promises two long-term advantages. One is lower costs: lower capital expenses because of higher utilization, less cabling, and fewer network connections. More significantly, it promises lower operating costs, from less labor if it's easier to automate data center management and have one infrastructure team as opposed to separate ones for storage and network management.

The second advantage comes if the unified infrastructure also increases IT agility by virtualizing IP and Fibre Channel storage networking and permitting management through a single console. No longer would new applications or changing capacity or an acquisition require separate cabling assignments and connection requests from the storage and network teams. A unified data center infrastructure promises to move companies much closer to the utility model of having piles of processing power and networking bandwidth and storage that they can divvy up as needs change.

However, a unified infrastructure will have to get past today's toughest measure: positive cash flow in a matter of months, not years. While vendors aren't advocating a rip-and-replace strategy, moving toward a unified infrastructure will require new hardware, and any hardware investment faces close scrutiny of late.

So why is unified infrastructure becoming a key discussion point now? One reason is that Cisco and rivals are making the case for it. Warrior's blog post generated a stir because Cisco is known to be developing a system, code-named California, that melds its switch for converging storage and data network traffic with blade servers and virtualization management. Cisco is expected to detail its unified computing strategy this week. That likely would put Cisco in competition with partners, such as IBM and Hewlett-Packard, that sell servers. Warrior acknowledges in a recent video that "in a consolidating market, you'll see the leaders of this industry compete in new ways."

Another reason unified infrastructure is gaining interest is that the unabated march toward virtualization is causing its own problems. Data center architects need more connectivity into virtualized servers because they're used for multiple purposes, as opposed to the days when a server might run only one application and thus have limited and static connectivity needs. Architects also are stumbling over the added complexity of managing physical and virtual layers separately, in addition to the silos of storage and networking.

A final factor spurring unification is a new Ethernet standard, close to being ratified, that's allowing for the lossless routing of storage traffic across Ethernet.

In the short term, view unification for what it is: a cost-containment strategy that also can simplify management, though most of that simplification is at the physical level, not the logical level.

But the real value is in operational flexibility. If the vendors deliver on their visions for a unified, virtualized infrastructure--management of the logical infrastructure becomes as unified as management of the physical infrastructure--it would let a single IT team run the entire network, monitoring and managing changes to both storage and standard IP traffic using a single console.

(click image for larger view)