We're not just offering software routing; we're offering open source routing. All routers contain some amount of software that handles the control plane -- running the routing protocol. There's also the forwarding logic, which is done in software and sometimes done in hardware. As you go up in performance, [the forwarding logic] is always done in hardware. At the very high-end carrier core, you always have something where the software routing protocol, OSPF or BGP, is running on a processor and the forwarding plane is running on an ASIC (application-specific integrated circuit). As you move down in performance to the midrange, you end up where the forwarding plane and the control logic running the routing protocol are done on the same processor. At one end of the spectrum, you have Linksys, where it's a little bitty processor, some memory and a couple of network connectors; and at the other end of the spectrum, you have a Juniper M160, which is a huge thing that has multiple ASICs doing forwarding and processors running protocols.
It's really a cost-performance tradeoff as you shift between those -- and complexity. Now, in the initial incarnation Vyatta's coming out with, we're focusing on delivering software-based routing on 32-bit X86 PCs (standard commodity hardware). There's no reason that it has to stay that way, and we actually wrote a
Linux came out in a very similar sort of circumstance; it was really targeting single-processor, X86-only PCs when it first arrived. Over time, Linux has broadened to run on everything from your cell phone to your supercomputer. And so there's ultimately nothing that says Vyatta has to be a software-based forwarding plane. You could run Vyatta and couple it with a hardware ASIC that would compete with the very high end. Or you could scale Vyatta down and run it on more embedded processors to compete with the very low end. What closed-source routers would Vyatta replace?
Initially, we believe that the x86 PC running Vyatta -- given the range of hardware that's available in the PC world -- can basically replace the midrange of the router market; to use Cisco terminology and model numbers, simply because it's convenient shorthand, basically from the 2800 series to the 7200 series. There's a whole host of equivalent products from Nortel and Alcatel -- but essentially in that range. I wouldn't describe it as Cisco model numbers so much as T1 branch office to gigabit LAN product categories. What are the pain points of those routers?
The biggest thing is really loss of control of the product. You don't really control the obsolescence cycle on your hardware; you're at the whim of a large vendor. To a certain extent, closed source is a lot like Stalinist central planning. It's a company saying that they know what's best, and they're going to deliver to you what you need regardless of your choice in the matter. Open source is a lot more free and open, and at the end of the day, really a lot more capitalistic, the way I think about it. It's the power of the ecosystem versus the power of central planning. I think there's a class of customer that chafes at being force-fed and really wants more control. What would be an example of that situation?
If you want to add something to your Cisco router, you're out of luck -- it's got the features Cisco wants it to have, and you're done. If you want to run a little software agent so you can monitor your Juniper router, you're out of luck because Juniper's decided what the feature set is going to be and you don't have any choice. If you want to run an old Cisco or Juniper router 10 years from now, you may or may not be out of luck because those guys may have "obsoleted" it and may not offer support on it anymore, basically forcing you to upgrade. Cisco seems to be expanding its product strategy beyond just networking devices and into all areas of IT networking. Does that leave a gap? Are they going to leave network pros out in the cold?
It seems to me that Cisco is, to a certain extent, a victim of [its] success. The good news is that throughout the 1990s, they were very successful in the networking market. The bad news is that now they need to go find other markets to sustain the growth rates that they had in the '90s. Briefly, how will your subscription model work?
It's an all-inclusive subscription model, with two tiers initially. There's a basic low-end tier for folks who have less critical installations and want guaranteed support but don't have as many time-critical requirements; then there's a higher tier that includes phone support and tighter guaranteed response times. It may be a small office or somebody that has a lab router, internal router or something internal that's not as critical who'd get the lower tier of support. Somebody who's got a more mission-critical installation would use the higher tier. I think we'll end up coming out with another tier later as we scale out our organization and we're able to offer better service to our customers. Often that comes down to a guaranteed response time and having the staff to cover things in a way that the customers want to see. What is the experience of using Vyatta from the perspective of the network engineer? Is it easy to integrate?
Our goal is to make it as seamless as possible, delivering something to the network engineer that he knows and expects. Today, that is a CD ROM that he can install on an X86 PC, which is a little outside the norm. Over time, we plan to offer appliances that will be prepackaged with software and hardware so that somebody can pull the system out of a shipping crate, rack it into a telco rack and start configuring it. That's a little bit more of the experience that a network engineer is used to over time, but other than that difference today, I think it's very complementary or compatible with the experience that a network pro is used to. My observation is that perhaps networking pros are less concerned about evangelizing open source and more concerned with finding technology that does the best job for the least cost. Do you think that is true?
I don't know that it's true that network engineers are not as evangelistic about open source. I think that they have not had anything to evangelize until now. To a certain extent, this bandwagon -- open source networking as opposed to open source server operating systems -- is later coming out of the station (to mix a metaphor). It's just not as advanced. If you wanted to compare them, you'd have to go back to Linux circa 1995 in terms of the maturity of the category of the market segment. While it seems there's a ton of open source advocates in general, I think that open source networking is kind of a new category. What is driving open source networking in the enterprise?
I think as we grow it and mature it, you'll see a ton of new open source advocates develop [around networking] in the same way that happened around Linux in server operating systems. One of the neat things about open source networking is that it allows a lot of other folks to get involved, which is great in terms of growing the community and the set of complementary solutions around a given technology. One of the things that customers see as a benefit is not only that they have a greater stake in the technology, but that the whole community starts to have a greater stake in the technology. Is there an open source networking technology that has already taken off?
What's been interesting for me is watching the Asterisk technology grow up. When it first came out, it was really just Asterisk. It was open source and you could get [it] and … use it. Now, what's started to happen is a lot of companies have started to announce Asterisk add-ons that really complement the base product.
That's a neat sign of things really developing and people seeing value and actually basing business plans on complementing an existing technology. In the closed-source world, that's a lot more difficult, because to build something onto Cisco's products -- if it requires looking at Cisco's source code or understanding the details of a proprietary Cisco protocol -- you have to sign a lot of legalese with Cisco to get that done. In the open source world, all the source code is there, so there's no need to establish a formal relationship with a given company. That allows things to really race quickly. The rate of innovation is much higher than it would be otherwise.
Note: Vyatta's open source routing software is available for free download from the company's Web site. Additionally, Vyatta is now offering tiered software subscriptions, including maintenance, upgrades and software support services, for the OFR. The Professional Subscription is available for $497 and the Enterprise Subscription for $647.
This article originally appeared on SearchNetworking.com.