Cisco Subnet An independent Cisco community View more

Cisco, Juniper, Arista: The data center networking stories just keep coming

Investigating what could make the "perfect" networking solution for the data center.

Meet Daye DeCentah, IT executive. Let's call her "DC" for short. DC faces a lot of challenges at work. Most all of these problems revolve around her having to deliver more, while using less - less staff, less systems, less power, less space, and, ultimately, less money. DC is bombarded with possible solutions to her wealth of problems. All have their advantages. All present both savings and improvements. All disrupt DC's flawed, but functioning, current environment - some more so than others. And all carry risks... now and into the future.

These last two weeks have been both wonderful and troublesome for DC. Many leading data center networking players have all pronounced their new solutions as the best. Who is telling the truth? Well... Everyone. And no one.

As every police detective will tell you, when hearing multiple stories describing the same event, the actual truth most likely lies somewhere in the middle. In the case of a flurry of a vendor pronouncements focused on the same network segment or service, the "perfect" solution likely lies somewhere in the middle. Unfortunately, DC cannot simply take a piece of Vendor C, and a piece of Vendor J, and a piece of Vendor A...and on and on.

Like a detective processing many different stories (and claims) from multiple witnesses and suspects, we are all left to form the best possible "truth" for ourselves. For me, I'm starting with certain basic beliefs - and moving forward with my investigation (and supplier interrogations) with these beliefs in hand. Here are my staring points...

The software is more important than the hardware - and it's not even close! It's a software world in networking. From high-level control functions to lower-level forwarding functions, it's the software that provides for the intelligence and dynamics that tomorrow's networks demand. It's the software that sets the network, the network operator, and the networked operator free in the future. That is not to say that hardware doesn't matter. It does. Hardware advancements boost network performance and reduce network costs. But hardware wins are fleeting. And let's face it, how many data center networks operate at the maxims laid out by this month's biggest, fastest networking hardware platform? For my "true" story, I'll give more weight to software that drives improvements in network simplicity, efficiency, and programmability.

The data center is one critical piece in a much bigger network puzzle. When did the data center network become an island unto itself? While certain network requirements in the data center demand special attention (e.g., super speeds, mesh designs, east-west flows), improvements focused on automated operations, resource utilization, threat mitigation, application optimization, etc. all have strong potential for impact across the entire network - campus and branch, LAN and WAN, wired and wireless. For my "true" story, I'll give more weight to data center advancements that can and will extend beyond the data center.

The network needs to adapt to your applications, not vice versa. While I applaud and support the continued movement to establish standardized northbound APIs, rewriting applications and retraining programmers is a more significant challenge than most people realize. IT organizations will likely only be able to adapt a select number and type of applications for network optimization. For the bulk of applications, they will be at the mercy of the underlying network. For my "true" story, I'll give more weight to the network that can do detailed analysis of application-specific loads and network state and take action on that analysis, all in real time.

The open networking components should be core to the solution - not a complement. Open systems are the best protection against all eventualities. And as we all know, all eventualities are in play when we're taking about IT infrastructure. Requirements come and go. Technologies come and go. Products come and go. Even suppliers come and go. Your organization -- and those it serves -- cannot come and go. Business shifts, process improvements, new revenue opportunities, enhanced service offerings...all and more can drive dramatic change to the IT infrastructure. And this change always needs to happen yesterday. For my "true" story, I'll give more weight to the solution that can be readily and rapidly extended to include new technologies, capabilities, applications, cost savings, etc.... from any supplier of choice.

The partner list is interesting, but not vitally important... at least initially. Every data center networking announcement from the last two weeks included lengthy lists of technology partners. So much so that many of the partner lists shared many of the same names. It's not the initial list of particular partners that is important. It's the combined solution and support that result from a particular partnership that is important. Partnerships almost never deliver on their full potential. It turns out that attracting a partner is a baby step compared to the long journey required to deliver a robust integrated offering. For my "true" story, I'll give more weight to the partnerships that focus on depth over breadth -- e.g., resource commitments, formalized interactions, firm roadmaps, and solution delivery goals.

Let the investigation begin!

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Must read: 10 new UI features coming to Windows 10