Moving to Fibre Channel over Ethernet in the data center

High data center demands require an efficient network to move data efficiently. Learn why Fibre Channel over Ethernet may be a good networking technology option, and how to make the switch.

As new technologies place heavier demands on the data center, organizations are being forced to scale up underlying technologies such as processing power, storage and infrastructure – all of which depend on an efficient network to move data efficiently. Over the years, a variety of communications technologies have come into vogue, ranging from ATM to InfiniBand to FibreChannel. While each promised important benefits, leveraging any of those technologies often entailed disruptive change in the data center and increased costs significantly. Today, data center operators are slowly considering the latest technology, Fibre Channel over Ethernet (FCoE).

Considering FCoE
FCoE also promises lower operating costs, ease of deployment, scalability, low latency and high performance – but these promises have been heard before. Still, FCoE may be different this time. It is based upon a very familiar communications technology -- one that is found in every data center -- Ethernet. Detractors may claim that FCoE is an attempt to save Fibre Channel by moving it to Ethernet. But FCoE is not about standardizing storage on Fibre Channel -- it's about standardizing on Ethernet as a data center network transport.

More on Fibre Channel over Ethernet:
FCoE network roadmap

Fibre Channel over Ethernet implementation

FCoE data center

Five tips on moving to Fibre Channel over Ethernet

Nonetheless, the jump to FCoE is still filled with challenges and potential pitfalls. Data center professionals must consider if it's worth the move, if FCoE has the potential to handle all data traffic, the amount of reengineering needed to deploy it successfully, and perhaps most important, the true cost of FCoE. To understand the implications of a move to FCoE, one has to understand how the technology works and how it will interact on the network.

The technology behind FCoE
Fibre Channel owes its success to the simple fact that the pain of existing storage networking technologies surpassed the costs and problems associated with deploying new technology (Fibre Channel, in this case). Will FCoE follow the same pattern? In one word, yes.

For example, FCoE encapsulates Fibre Channel frames over Ethernet networks. That allows Fibre Channel to use 10 Gigabit Ethernet networks while preserving the Fibre Channel protocol. FCoE brings the ability consolidate fabrics, reduce cabling and leverage existing Fibre Channel networks as well as existing management software – some significant benefits for storage across Ethernet networks.

FCoE relies on two rule sets based upon newly defined industry standards. The first set of Fibre Channel rules is based upon the T11 FC-SW standard for handling Fibre Channel frames. The second set of rules is based upon the IEEE 802.1D standard and handles how Ethernet frames move across the network. Collectively, those standards map Fibre Channel frames onto Ethernet frames and create a reliable Ethernet transport mechanism.

Deploying FCoE
Although vendors would like customers to believe that FCoE adoption will be a "plug-and-play" affair, it won't be that simple. There are several prerequisites and supporting technologies that need to be addressed before moving to FCoE. The basic components required for FCoE include converged network adaptors (CVAs), which combine the functionality of Ethernet NICs and Fibre Channel host bus adapters; Fibre Channel over Ethernet switches; and new optical or coaxial copper cabling -- FCoE does not support RJ45 connections.

This might mean waiting for a server and infrastructure refresh before making the jump -- new servers, racks and switches may come with FCoE support already integrated. Ultimately, every element of the data center will need to be checked for compatibility. For example, some rack-based servers may not have the room or capacity to support a CVA, or the racks in place may lack room for an FCoE switch.

Organizations already using Fibre Channel will typically find implementing FCoE somewhat easier, since familiarity with the technology already exists and management software will be in place. However, that is not a requirement. In most cases, the move to FCoE will not require a complete re-engineering of the data storage system. This is important since it allows FCoE to be deployed incrementally, while running side by side with traditional Fibre Channel deployments.

One of the most critical elements here is the ability to collectively manage existing Fibre Channel deployments side by side with new FCoE deployments. A unified management system and administrator-facing dashboard will prove to be the primary methods for judging health, performance and creating reports that show efficiency. Without an integrated management toolset, administrators will have to turn to multiple management consoles that cannot correlate events with traffic, usage and status of the given arrays. One of FCoE's primary promises is to offer unified management across legacy Fibre Channel systems and new FCoE implementations. That said, due diligence requires that management utilities be validated and tested in hybrid environments to guarantee that storage area network (SAN) management will not become a problem for enterprise storage administrators.

For most organizations, a migration to FCoE starts with connectivity between the servers and top-of-rack switches (the access layer). A move further into the network infrastructure will come later, once FCoE is vetted and established in racks. Initially, FCoE will affect SAN and related servers, but FCoE does not directly integrate servers and storage arrays. That means purchasing new FCoE switches for top-of-rack deployment from a major vendor, such as Cisco, Brocade, NetApp or EMC. Those switches split the Ethernet-based LAN traffic and the FC-based SAN traffic, allowing users to tap into their existing core switches and FC SANs.

This is currently the safest way to move over to FCoE -- top-of-rack switches allow the existing FC infrastructure to stay intact, still running from the core switches to the disk arrays. That methodology will also offer cost savings, since most of the benefits realized will come from the access layer between the servers and the first-hop switch. So an organization can benefit from FCoE technology without immediately moving over to an end-to-end FCoE to realize any savings.

Increments and vendor support
While the benefits of FCoE are clear, organizations will want to take an incremental approach to deployment, leveraging falling prices for the technology and benefiting from the information and expertise garnered from pilot projects done elsewhere.

And don't hesitate to leverage the vendor for support. Most FCoE vendors offer case studies, white papers and expert advice that is sure to ease the move to FCoE. Enterprises will want to develop roadmaps that combine hardware refreshes with FCoE deployments and use vendor support to guarantee success, without encountering any unforeseen problems.

What did you think of this feature? Write to SearchDataCenter.com's Matt Stansberry about your data center concerns at  [email protected].

Dig Deeper on IBM system z and mainframe systems