Optical/IP Networks

Laurel Launches Edge Router

Laurel Networks has announced its first product: the ST200 edge router. The ST200 is designed for installation at the edge of an IP/MPLS network. Laurel’s biggest competitor in this space is Cisco Systems Inc. (Nasdaq: CSCO).

Taking on Cisco at routing would generally be considered a foolish move for any small startup, except for one thing: Laurel’s strategy is remarkably similar to that employed by Juniper Networks Inc. (Nasdaq: JNPR) -- the only company that has successfully challenged Cisco’s dominance of the router market to date.

The key difference is that, whereas Juniper attacked Cisco in the market for core Internet routers, Laurel is focusing its efforts at the other end of the network: the edge.

But has the ST200 got the right stuff? The product boasts an array of features optimized for the edge of the network, including high-performance distributed routing and forwarding functions, as well as dedicated -- or “per-customer” -- virtual routing tables, traffic shaping, class-based queuing, and differentiated services. Additionally, the device will use hardware-based counters to collect billing stats, Laurel says.

The ST200 also comes with “multi-lingual” line cards, supporting multiple protocols. This allows service providers to reduce capex by using the same hardware to route ATM, frame relay, gigabit Ethernet, and IP traffic over the same backbone -- without having to buy and swap different flavors of card in and out of the router chassis.

“[The ST200] is interesting, design-wise,” comments David Newman, president of Network Test Inc.. “Virtual routing allows you to dedicate router resources to each customer, and that’s a very smart move -- especially if you are making boxes that are customer-facing. The big challenge for Laurel will be keeping the content of those tables in sync. If you have N tables, how do you ensure that table 1 has the same contents as table N? That can be a real showstopper.”

Newman says that Laurel faces an uphill battle competing with the likes of Cisco and Juniper, but not an impossible one. “Edge routing is a new market. Cisco is dominant, but not necessarily indomitable. Can a startup make money at its expense? It’s possible, and I’d offer Juniper Networks as exhibit A."

Still, it won’t be easy. Cisco isn’t resting on its laurels (sorry). It’s already made some pre-emptive moves to strengthen its edge routing story (see Cisco Puts Service Creation on Edge). Juniper’s product line is starting to look increasingly edgy (see Juniper Goes to the Edge). Unisphere Networks Inc. also has recently started to make some serious money from sales of its ERX edge router (see Unisphere Posts 47% Revenue Growth ). And several other startups are targeting this area, including Amber Networks Inc. (see Amber Illuminates Edge Product), and Quarry Technologies Inc. (see Quarry Mines Another Product).

Laurel says the ST200’s biggest differentiator against the products from its competitors is speed. “We’re not trying to take folk on in the T1 market. We’re more of a high-end edge play,” says Stephen Vogelsang co-founder and vice president of marketing at Laurel. While the ST200 will support T1 (1.5 Mbit/s) connections, where it really excels is supporting very large numbers of DS3 (45 Mbit/s), OC3 (155 Mbit/s), OC12 (622 Mbit/s), and gigabit Ethernet ports, he claims.

Laurel makes no secret of the fact that it is trying to follow in Juniper’s footsteps, contending the strategy has helped it secure financing. “That’s exactly what we told the VCs, and it made it a lot easier to get funding,” says Vogelsang. The company has garnered almost $80 million in two rounds of funding (see Laurel Networks Scores Big Round ).

Still, there’s one important area in which Laurel is not following Juniper’s playbook: the poaching of Cisco employees. Juniper made a point of hiring large numbers of Cisco software engineers to ensure that the code in its routers would be compatible with Cisco’s. This turned out to be a key move. It increased service providers’ comfort level with Juniper’s products and made it much easier for them to deploy its routers in heterogeneous networks.

“I don’t know that we have anyone from Cisco on the development side,” says Vogelsang. “Then again, there may not be anybody left there to take.”

Laurel says it will demonstrate the ST200 at the Supercomm trade show in Atlanta next month.

— Stephen Saunders, Founding Editor, Light Reading

For more information on Supercomm 2001, please visit the Light Reading Supercomm 2001 Preview Site.

Page 1 / 2   >   >>
Belzebutt 12/4/2012 | 8:23:39 PM
re: Laurel Launches Edge Router "The big challenge for Laurel will be keeping the content of those tables in sync. If you have N tables, how do you ensure that table 1 has the same contents as table N? That can be a real showstopper.GÇ¥

Why would you need ensure that all tables have the same content? The whole point of this virtual router capability is that each customer can have a different routing table, simulating their own router.
dnewman 12/4/2012 | 8:23:33 PM
re: Laurel Launches Edge Router Er, um, good point. I misspoke.

Multiple tables aren't identical as far as customer info is concerned.

However, the same challenge applies -- it's still very tricky to keep the *rest* of the routing tables in sync. This is true even if there's only one common entry across all tables.

This problem has bedeviled several past efforts at building virtual routers.

David Newman
Network Test
kupfi 12/4/2012 | 8:23:30 PM
re: Laurel Launches Edge Router I read somewhere that Gotham Networks is doing Virtual Routers too. Are both in the same space?
dnewman 12/4/2012 | 8:23:27 PM
re: Laurel Launches Edge Router "Each VR is managed independantly. So whether or
not there are common entries in different VRs
is really an orthogonal issue. "

First sentence true, second sentence false.

Devices with multiple tables propagate routing updates across the tables, even if they are managed independently. If (or, more likely, when) the system scheduler gets hit with enough other tasks to get bogged down, the result can be significant time lag between an update for table 0 and one for table N. Hence, the routing may be all wrong for the customer relying on table N's contents.

Note that I'm not saying this is a problem with Laurel's or any other specific implementation. In the past has been an Achilles heel for some versions of this architecture. It will be interesting to see in trials how well Laurel has worked around the issue.

David Newman
Network Test
alligator 12/4/2012 | 8:23:25 PM
re: Laurel Launches Edge Router Both the Laurel and the Crescent Networks announcements speak about the virtue of virtual routers. Given shared processing and the problems mentioned by Dave Newman, how can a carrier possibly offer any SLA guarantee to an enterprise customer? (Ok Ford - see that processor? it is calculating your routing and the routing for 1000 other customers) I'd contend that this is the problem that has hampered VR deployment all along, and that although VRs are a good marketing ploy, they do not hold up under actual deployment. Is there any hard evidence to the contrary?
skeptic 12/4/2012 | 8:23:23 PM
re: Laurel Launches Edge Router
"virtual router" can mean different things.
You have to carefully look at what the company
is selling and what their model is for making
use of it.

There is an oversubscription problem in any router. And if someone wants to use virutal
routers, they have to sit down and figure out
how many they get per box and how much they have
to change for the service to make a profit.
Its like any other service on the edge.

The problem isn't necessarly the calculation time for routing, its that routing has a "footprint" in memory (a fixed cost) and virtual routers
tend to eat up memory because of the replication of the footprint. But again, it depends what they mean by virtual router.

dnewman 12/4/2012 | 8:23:21 PM
re: Laurel Launches Edge Router Hey, I *like* the idea of virtual routers! They offer some nice advantages, like the ability to give multiple customers the same net-10 blocks without having to deploy multiple routers.

As skeptic notes, there are different approaches to VR design. I frankly don't know enough about the new entrants' boxes whether sync is even a problem in the first place. It might be, it might not be. Let's see what turns up in trials.

David Newman
Network Test

packets_circuits 12/4/2012 | 8:23:17 PM
re: Laurel Launches Edge Router Different VPN techniques (such as MPLS VPN)
also allow different customers to use
overlapping address spaces etc so what special
capability is enabled by virtual routing ?

In fact one can see several advantages of the
MPLS VPN approach over a VR based one ...
netskeptic 12/4/2012 | 8:23:16 PM
re: Laurel Launches Edge Router > In fact one can see several advantages of the
> MPLS VPN approach over a VR based one ...

Interesting, I know one: this is the only way CISCO can immerse itself into VPN market place.

I would even venture to say that the whole (G)-MPLS has only two advantages: (1) see above, (2) CISCO and JUNIPER can buy another 3 years while BGP based systems will be still viable in the core.

Again, just an IMHO.



gladysnight 12/4/2012 | 8:23:15 PM
re: Laurel Launches Edge Router "In fact one can see several advantages of the
MPLS VPN approach over a VR based one ..."

Great. Except that you resolutely fail to mention any.

Care to offer some?
(I seriously want to know, the above is just a gentle bit of needling. Can't help myself.)
Page 1 / 2   >   >>
Sign In