Welcome!

Open Source Authors: Adrian Bridgwater, Liz McMillan, Elizabeth White, Pat Romanski, Roger Strukhoff

Blog Feed Post

From white box to bare metal: Dell to run Cumulus SW on Force10 switches

Big news in networking today as Dell announced plans to ship its Force10 switches with Cumulus software. The news is covered in several places, including the following:

These articles do a good job of explaining what Dell and Cumulus are up to. I won’t rehash the product or technical details here. But what are the implications of this announcement?

When Cumulus first came onto the networking scene, the talk was all about decoupling networking hardware and software and how that would help lead the white box switching movement forward. The hope was that by separating the brains from the brawn – the network OS from the underlying switching hardware – the switches themselves would become dumb transport devices, allowing them to be replaced by commodity switches.

At the same time, they advanced a familiar thread in networking by saying that the decoupling of hardware and software would help fight vendor lock-in. At the time, this argument specious. Supporting a network OS across a set of functionally identical white box switches doesn’t provide any choice. Customers would basically have their choice of the same device manufactured by several different vendors. Choice of manufacturer is not really what people were looking for, so the optionality argument put forth was somewhat disingenuous at the time.

Today’s announcement changes both of these points in a significant way.

By shipping their Linux-based networking operating system on a mainstream vendor switch, Cumulus is proving that they are capable of doing more than just running on top of a reference switching design. This means that customers have the option of selecting into a Cumulus environment and effectively swapping out the underlying devices depending on the specific need. That is not to suggest that there are not barriers to change once you get settled on a single platform, but those barriers are more palatable when the software that runs on top remains constant. Indeed, much of the lock-in for any networking solution is in the software. If the software is constant, the hardware is more open to change.

But the subtle point here is that the driving decision for customers might not always be the lowest possible price for a hardware platform. If the future of networking is going to be the lowest possible CapEx, then the only platforms Cumulus ought spend any time supporting are the lowest-cost platforms. But that’s not what’s happening.

To date, Cumulus has been somewhat complicit in propagating the message that it is all about CapEx. They have driven, or at least allowed, the conversation towards CapEx. Strategically, this was a pretty crafty move. The story, even if not entirely true, is pretty easy to understand, and if there is one thing that we have learned with this whole wave of SDN and related technologies, it’s that people are moved by simple stories. There is no simpler story than saying that the future of networking will necessarily follow the server evolution, resulting in cheap white box switches.

Somewhere along the way, the story started to change. White box became bare metal, and, if you watch closely, CapEx is becoming OpEx.

This was a necessary shift. The reality is that competing on a pure CapEx difference was never going to be a successful long-term play. While CapEx will work for a time, the simple fact that most of the switching world is converging on a narrow set of switching silicon means that the difference in price between the low and high ends of the market is going to become smaller over time. On top of that, the high end of that space is dominated by incumbents who have enough volume to negotiate better per-unit pricing, which allows them to get even more competitive on price.

So what is likely next for Cumulus?

This should be a relatively noticeable inflection point in their go-to-market strategy. I would guess that 2014 becomes a lot more about OpEx than CapEx. They should start to lean more heavily on their Linux foundation (though, to be fair, almost everyone has a Linux base at this point). They will likely shift their public dialogue more towards automation and DevOps. In private meetings, they will likely talk about support pricing as well (particularly in comparison to Cisco support costs, a long-time customer pet peeve). I say that the supporting pricing will be private because any real public discussion about pricing allows Cisco to just change their support costs on a per-deal basis and effectively squash that differentiator.

The interesting impacts, though, are not really related to product. Now that Cumulus has lined up Dell, how long before some of the other vendors come along? It is not a huge stretch to see someone like Extreme come along for the ride. They already announced support for Big Switch’s SwitchLight software, so opening up to another player would be a natural move (though it would depend on the terms of that relationship, obviously). After that, HP becomes interesting. How far up-market can this go? When all is said and done, can Cumulus hop on board enough single-digit players to move their addressable share to north of 10%? It is worth noting that Cumulus will not be sold on 100% of their partner solutions, so it really is addressable share more than actual share.

And if they do make progress over time, what happens to the industry if VMWare takes them out? It could be that Cumulus is an interesting way to add a software-only product to the networking portfolio, a nice complement to NSX. This would basically galvanize a number of players against Cisco, allowing a common go-to-market framework. Sure, the individual players lose a bit of their ability to differentiate in software, but market share hasn’t been moving for many of these companies anyway. Their best bet could be to become a nuisance to Cisco in switching, which could tie Cisco up in a way that makes them less capable of responding on other fronts.

The companies that become extremely interesting in this scenario are those companies who have made gains in switching. Brocade has been making progress, so it would be more difficult for them to make the leap to a Cumulus network OS like the others. Brocade has already started talking OpEx and automation, so they could have the go-to-market capabilities to go it alone, drafting on the shift in industry dialogue.

It’s difficult to predict what will happen with any precision, but this partnership certainly has war rooms across the industry spun up today. Ultimately, these types of moves are good for customers, even if they make life a little chaotic on the vendor side for awhile.

[Today's fun fact: A Boeing 747′s wingspan is longer than the Wright brother’s first flight.]

The post From white box to bare metal: Dell to run Cumulus SW on Force10 switches appeared first on Plexxi.

Read the original blog entry...

More Stories By Michael Bushong

The best marketing efforts leverage deep technology understanding with a highly-approachable means of communicating. Plexxi's Vice President of Marketing Michael Bushong has acquired these skills having spent 12 years at Juniper Networks where he led product management, product strategy and product marketing organizations for Juniper's flagship operating system, Junos. Michael spent the last several years at Juniper leading their SDN efforts across both service provider and enterprise markets. Prior to Juniper, Michael spent time at database supplier Sybase, and ASIC design tool companies Synopsis and Magma Design Automation. Michael's undergraduate work at the University of California Berkeley in advanced fluid mechanics and heat transfer lend new meaning to the marketing phrase "This isn't rocket science."