IP & Optical Take Different Roads Toward Integration
I spoke with Heavy Reading senior analyst Sterling Perrin recently about issues expected to affect the optical transport sector in 2015. Part 1 of our conversation focused heavily on metro 100G. Here's Part 2, in which Perrin digs into the integration of IP and optical at both the hardware and software layers.
Light Reading: IP+Optical integration is becoming an increasingly popular topic. But what are we really talking about?
Sterling Perrin: There are two pieces to it. There's the physical integration of the IP and the optics in the same piece of hardware, which is similar in concept to what has been talked about for a decade-plus with IP plus DWDM and IP over glass concepts. The newer piece, which is driving a lot of the recent interest, is the management and control plane integration of the IP and optical layers, which a lot of operators are interested in now. Along with the coming of SDN, and its promise to make things open rather than proprietary, operators will have a lot of opportunities that were closed to them with the proprietary hardware integration.
Light Reading: How is the industry coming along with software layer integration?
Perrin: I don't see a lot yet being done by the operators. There are some road blocks. The big road block when you talk to the big network operators is that they really need things to be fully open and standardized. In our recent webinar, the Telefónica story in general was that they are really trying to drive a lot of that standardization of the software control between the two layers. Years ago, they developed software of their own that could glue products together, but the problem was that they had to do all that work themselves and it was specific to the specific vendors they were using and the specific releases of those vendors. So, the problem that Telefónica has, and really any large operator trying to do things at scale, is that if you don't have standards, then every time you change vendors or every time a vendor comes out with a new product or new releases, you're going back to the drawing board and reworking all of your software that's integrating those layers.
Light Reading: And that has some impact on scalability, right?
Perrin: What it means is that you can't scale efficiently, and if you can't scale efficiently, there is really not much reason to do it. The challenges outweigh the benefits, so it really needs to be open and standardized. Even though operators have historically done things on their own to glue their own vendors together, that's really not the end game this time.
Light Reading: Is this where SDN comes into the mix?
Perrin: That's why SDN is so important. Where we are right now is that different operators and different vendors are coming up with different means to integrate the layers with software, and putting them in standards bodies, and the standards bodies are working with them and trying to come to agreement. We're probably still several years away from a point where this is shaken out and the industry has a clear roadmap for standards. Transport SDN is a subset of carrier SDN. It's interesting that a lot of the carrier SDN discussion so far has been dominated by the optics guys. Operators are more interested in SDN driven by the IP guys. SDN benefits are primarily in Layer 3, and if you're viewing optics in isolation, it's way down at the bottom of why operators want SDN.
Light Reading: Is physical integration of IP and optical going any better than at the software level?
Perrin: The physical integration is an easier story because if you buy into a vendor's box that has that physical integration, that's the end of the story. But this has been slow to move. There was a lot of talk a year ago, even two years ago, from vendors about boxes that would integrate, but it's been hard to pull from these guys what's actually been bought and sold and deployed. There is still interest in the physical integration, though. If you look at IP and optical in the metro market, it really does seem like operators and vendors are starting to integrate the two physically, but the numbers are hard to get at. It's still fairly new, and not a lot of it has been done yet. Potentially, over time, that could be a good opportunity as well.
Light Reading: What are a few examples of how vendors are doing physical integration?
Perrin: Juniper has the PTX 3000 for metro, which they've integrated with optics. Cisco has integrated optics on its edge routers. Then they've got their next-generation NCX 4000 product, which is also does the same sort of integration, integrating packet from the transport side. Alcatel-Lucent has talked about and introduced some things. Ciena has the 8700, which they came out with in the fall of last year integrating Layer 2 and WDM optics. So those are some of the products. There have been some announced and also some available, but in terms of what's being deployed, the market hasn't taken to them yet.
Light Reading: If software-based integration accelerates, will physical integration accelerate in parallel or slow down and become less of a priority?
Perrin: I think you have to look at the core and the metro a bit differently. The way forward for the core is going to be management and control integration. There doesn't seem to be a lot of appetite for physical integration there. You're dealing with massive capacity, and then there's distance. What's happening with 100G, with the amount of capacity that's needed and the distance requirements, you're seeing the client side density of 100G is better than the line side of 100G. To boost those distances, there's other stuff that needs to go into these 100G transponders, making them bulkier and costlier. Basically, it's a faceplate problem. You give up capacity in a router if you put line side 100G optics in it versus just putting it in the client side. So, you're wasting expensive, much-needed capacity. For those reasons, I don't think the core is going to be where the physical integration occurs.
Light Reading: And what about the metro?
Perrin: We're seeing specific metro 100G transponders with shorter distance capability because they eliminate all of the bulky electronics that are used to boost distances to thousands of kilometers for long-haul. We've seen the CFP form factor [smaller and lower power] optics going on metro boxes. It's the same form factor as the client side 100G, but it's also doing the line side, so you're not making that faceplate sacrifice. It may be that CFP-2 is the place where we see the physical integration occur. Operators might just be waiting for that CFP-2 wave to take hold because they don't want to waste capacity.
— Dan O'Shea, Managing Editor, Light Reading