Sun vs. Cisco? I’m Getting My Popcorn…
Scott Lowe wrote an interesting blog today wondering if Sun was preparing to take on Cisco in the virtualization space, referencing the development of virtualized networking functionality featuring the novel combination of commodity hardware and open source software to unseat the Jolly Green Giant:
A while back in Virtualization Short Take #25 I briefly mentioned Sun’s Crossbow network virtualization software, which brings new possibilities to the Solaris networking world. Not being a Solaris expert, it was hard for me at the time to really understand why Solaris fans were so excited about it; since then, though, I’ve come to understand that Crossbow brings to Solaris the same kind of full-blown virtual network interfaces and such that I use daily with VMware ESX. Now I’m beginning to understand why people are so thrilled!
In any case, an astute reader picked up on my mention of Crossbow and pointed me to this article by Jonathan Schwartz of Sun, and in particular this phrase:
You’re going to see an accelerating series of announcements over the coming year, from amplifying our open source storage offerings, to building out an equivalent portfolio of products in the networking space…
That seemingly innocuous mention was then coupled with this blog post and the result was this question: is Sun preparing to take on Cisco? Is Sun getting ready to try to use commodity hardware and open source software to penetrate the networking market in the same way that they are using commodity hardware and open source software to try to further penetrate the storage market with their open storage products (in particular, the 7000 series)?
It’s an interesting thought, to say the least. Going up against Cisco is a bold move, though, and I question Sun’s staying power in that sort of battle. Of course, with Cisco potentially distracted by the swirling rumors regarding the networking giant’s entry into the server market, now may be the best time to make this move.
It's really the last paragraph that is of interest to me, specifically the boldfaced sentence I highlighted. I think the "rumors" have pretty much been substantiated by the mainstream press, so let's assume "California" is going to happen.
- I don't know how anyone can think that Cisco is "distracted" by bringing to market the logical extension of virtualized infrastructure — the compute function — as anything other than a shrewd business decision to offer a complete end-to-end solution to customers. I talked about it here in blog post titled "Cisco Is NOT Getting Into the Server Business…" This is an Enterprise Architecture play, pure and simple.
- Honestly, if we're discussing commoditization, a server is a server is a server, whether it's in a blade form factor or not, and it's not like Cisco has to worry about building things from scratch. The availability of OEM/ODM components (raw or otherwise) means they don't have to start from scratch. Oh yes, I know HP spent a bazillion dollars on C-Class fan engineering and IBM's BCHT is teh awesome and…
- The whole game is Unified Computing; bringing together enterprise class compute, network and storage as a solution with integrated virtualization, management and intelligence; you take the biggest pain point out of the equation — integration — and you drive down cost while increasing utility, agility and efficiency.
- If you look at what "California" is slated to deliver it's hard to see how Sun would compete: A blade based chassis with integrated Nexus converged networking/storage, integrated virtualization from VMware (with Nexus/VN-Link,) and management from BMC. You know, Enterprise stuff, not integration hodge podge.
So, I ask, does this look like a distraction to you?
Hi Hoff,
There are just a few thoughts I'd like to share:
1. No organization, no matter how well it executes, can take on a new market–a market with deeply entrenched competitors, and a market that is radically different than their previous markets–without exposing themselves to competition in their core market(s). I don't disagree that what is projected to be included in "California" is a logical extension of virtualized infrastructure and might be a very compelling offering. But in the server market, margins are razor-thin, the competition is fierce, Cico's opponents are well-funded and have decades of experience over Cisco, and have a sales force that knows how to sell this stuff. It will take lots of resources and lots of time for Cisco to compete with HP, IBM, Dell, etc., and while they are focusing on that the risk for a loss of market share in their core market is there. That's what I mean by being "distracted" by "California". Make sense?
2. My mention of Crossbow–Sun's virtualized networking functionality–was less about Crossbow itself and more about just one more piece in a larger networking puzzle. But think about where Cisco is heading with NX-OS: virtualized instances that run certain parts of a router's functionality. Doesn't that sound like Solaris Zones? There's an interesting mix of technologies here, perhaps coming from a different origin but still heading in the same direction. Here's the product description: An operating system with virtualizable full-featured network routing, firewalling, load balancing. Which one does that describe? NX-OS or OpenSolaris?
3. I'll wrap all this up by simply saying: While Sun may be preparing to take on Cisco, I don't necessarily believe Sun will succeed. Sun may have a great mix of technologies, but it takes more than technology to succeed. Cisco has great execution skills and tremendous staying power, both of which Sun lacks. I'd love to see Sun succeed, but….I'm not holding my breath.
Thanks!
To your points, Scott;
1) My argument was that given the commoditization of "servers," and with virtualization leveling the playing field, it's really not that much of a "radically different" market for them in my estimation. It's a platform. It will be sold as a solution. I don't disagree that it's a market they haven't traditionally sold into (servers) but think of this as an APPLIANCE, which Cisco has sold for years.
It shouldn't be marketed or sold as a "server." It's unfortunate that people are calling it such.
Further, Cisco doesn't enter a market to be #2 or #3.
I totally agree that "servers" aren't a core competency for Cisco, but appliances and platforms are, which is what "California" will be.
2) I don't really understand your point here. Can you clarify what you mean?
3) Agreed. We'll see what happens next week when Sun's strategy changes again. Who knows, perhaps they'll bring back the Nauticus Switch and say they're a security company again… 😉
—
Again, I was picking up on the issue you raised (that others have commented on) regarding the "distraction" theme. I don't think they're distracted at all. I think they're right on target.
I originally predicted that we'd see apps running in the Nexus switches natively. "California" makes MUCH more sense, especially with VN-Link and the NX1Kv and what will follow.
Good points!
1) To a certain extent, you are correct. The rise of Intel Nehalem and its related chipset (QuickPath, I believe it's called?) will give rise to virtualization "appliances" from a number of vendors–HP, IBM, Dell, and Cisco. But aside from the potential integration with the Nexus technologies, what will differentiate "California" from any other Intel Nehalem-based server? Why should I treat Cisco's blades as an appliance, but HP's blades as servers when they are built from (essentially) the same underlying hardware? Further, you can call it an appliance all you want, but if the market perceives it to be a server, then it's a server.
I can tell you that from my interactions in the field, the fact that Cisco knows how to sell appliances that perform networking functions doesn't automatically translate into knowing how to sell appliances that perform compute functions. Again, you can call it an appliance, or a solution, but it performs a function that is fundamentally different than other things that Cisco sells, and that is a challenge that the Cisco sales force must overcome.
2) In your post you seemed to pick up on Crossbow in particular–sorry if I misread that. I just wanted to point out that Crossbow was one of several different technologies that were coming into play here.
3) I think we're in complete agreement here.
I think you and I are getting hung up on the word "distracted". When you read me talking about Cisco being "distracted," think of distracted as meaning "focused on something else". I'm not saying that "California" isn't a good move for Cisco, or that isn't a natural, logical extension of their product and vision, or that I think they're stupid. But, as you pointed out, Cisco doesn't enter a market to be #2 or #3, and to be #1 in this market it will take resources and focus. And while Cisco's focus is on "California," an astute competitor (OK, does that exclude Sun?) has the opportunity to make a move.
Thanks–great discussion!
Hello. Interesting discussion. I thought I'd throw in a few cents worth…
Regarding point 4: "If you look at what 'California' is slated to deliver it's hard to see how Sun would compete: A blade based chassis with integrated Nexus converged networking/storage, integrated virtualization from VMware (with Nexus/VN-Link,) and management from BMC. You know, Enterprise stuff, not integration hodge podge."
Sun does have most of those items. Blade chassis: Sun Blade 6000 chassis. Integrated networking: not that I know of, but one could imagine something based on their 10GbE chipset or their infiniband switches. Virtualization: xVM Server. Management: xVM Ops Center. I guess the big question is how well that'd compete with Cisco on both a usability and price point basis.
My trouble with Sun is that no matter how many of the components they own, the way they integrate them always seems like a hodge podge. However, I think this comes from making that integration optional to keep from locking the clients in, rather than VMWare and Cisco who are more interested in keeping you in their cage. Not that there are many alternatives when you consider some of the advanced things that VMWare is doing.
I still think that Sun is coming into the game at a level of integration that VMWare was at about 4 years ago.
This Pretty Much Answers That Question
A few days ago I asked if Sun was preparing to take on Cisco in the networking space. It looks like that question has been answered