In recent months, we've seen a wave of specialized blade servers aimed not at general processing of application workloads but very specific tasks, namely networking.
Some people are calling this "enterprise on a blade," but there are growing questions about whether this is the best use of blade technology. Sure, it might be economical, but is it optimal?
Some of the more recent examples of this trend include Brocade's new DCX-4S, the mid-level version of its DCX Backbone switch. As I mentioned in Tuesday's post, the unit offers up to 192 8 Gbps ports and is very adept at SAN partitioning to accommodate virtual environments.
Fujitsu has taken a similar tack with its NX650 Primergy NAS blade that lets you integrate data from the company's BX600 S3 chassis and then scale the system up to 3.7 TB of storage. The company says that if you are struggling to keep tabs on multi-system storage environments, consolidation onto a blade environment is the way to go.
But probably the leader in the blade networking trend is, not surprisingly, Blade Network Technologies, which relies on the platform for a range of GbE and 10 GbE systems. The company has started to venture into a range of advanced technologies for its switch platform, such as "migration-aware" technology that enables live migration across virtual data centers.
The idea of putting all these networking services on a blade is certainly compelling. It's usually cheaper than a dedicated hardware device, and lends itself to more simplified infrastructure and easier management, provided the data center in question is already well-versed in blade technology.
But again, the question arises: is it the most optimal solution? Steve Taylor, president of Distributed Networking Associates, and Larry Hettick, principal analyst at Current Analysis, have a pretty lively discussion going in their newsletter on this subject. Their main argument is that a blade can be configured to hold a lot of varied tools, but they might not be best-in-class. A Swiss Army Knife can hold lots of tools, too, but you wouldn't want to cut a steak with it or use the scissors to fashion a new suit.
There's also the question of rapid obsolescence, considering blades use general-purpose ASICs that may or may not be able to handle future protocols or software upgrades.
To be sure, these are valid concerns. But I think Steve and Larry hit it on the head at the end of their blog when they call for business solutions on a case-by-case basis. Not everyone needs best-in-class, and in a tough economic environment, it's likely that some formerly high-flying enterprises will come to realize that "good enough" might have to suffice for a while, with an eye toward future upgrades as budgets allow. After all, swapping out a blade with aging processors is supposed to be a snap, right?
There's a tendency in human endeavors to think there's a right way to do things, and a wrong way. If there's a larger trend at work here beyond the rise of blade-based networking, it's that enterprise technology is becoming so diverse that there are likely to be multiple right ways from now on.