On the other hand, server admins who are well versed in Linux and specifically Linux networking will feel right at home, after a fashion. They will have no problem navigating, making modifications, or scripting. But they may find themselves puzzled by certain networking aspects that aren't commonly used in the Linux server world, such as Quagga, the open source routing suite that brings OSPF, BGP, RIP, and other routing protocols to Linux. General Linux server admins are unlikely to have run across the need to deploy Linux-based routers in the past, and they will have to brush up on these complex protocols in order to deploy Cumulus Networks's switches.
So we'll potentially have one side of the house that knows the needs, and another side that knows the hows, while neither has the whole picture. Given the nature of networking and the fact that in many infrastructures high-level routing protocols are fairly static once configured and working, I might wager that the Linux admins will win this particular race -- simply because the interface is their bread and butter. Well, that and the fact that you can easily script widespread changes to Cumulus switches by hand, or with Puppet, or Chef, or whatever you like. It's just a Linux box, after all.
Another element to consider is the loss of a central configuration file. This has always been a strength of core networking devices made by Cisco, HP, and others. Because you can copy a single file off of a switch or router that contains the entirety of the configuration necessary to make that switch function, you can easily bring a replacement switch up in a very short time -- no matter how complex that configuration might be.
That's not the case with Cumulus Linux-driven switches. You might find configuration elements in many places within the file system, stored in individual configuration files and each with its own syntax. This really underscores the need for centralized configuration management for Cumulus Linux. If you have Puppet or Chef in place already, then you can deploy a Cumulus switch in about the same time as a "normal" Cisco switch. However, without that management infrastructure, the Cisco deployment will win every time, potentially by a large margin.
I think that in the short term, a lot of larger infrastructures will want to look very closely at Cumulus. They will already have the means to deploy lots of switches easily, along with the networking requirements for very fast switching. Smaller shops will be enticed by the pricing, but may balk at the implementation costs if they don't have the necessary skills to bridge the Linux and networking gaps or because they don't have centralized configuration management in place.
Either way, Cumulus Linux puts pressure on every established switching vendor to either lower its prices or lose sales -- some of which will be extremely large in quantity. The divorce of switching software from switching hardware will not be amicable, nor will it be speedy, but I think it's destined to occur. It might be time to brush up on your Linux or your routing protocols just in case.
This story, "Move over, Cisco IOS: Dawn of the industry-standard switch," was originally published at InfoWorld.com. Read more of Paul Venezia's The Deep End blog at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.