x
Routing

AlcaLu Tunes Up Redundancy

Alcatel-Lucent (NYSE: ALU) is beefing up high-availability features on its switches and routers, playing up one of the features that's helped it gain ground against Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (NYSE: JNPR).

High availability generally involves having backups available -- a second switch fabric that can take over after the primary fails, for instance. Alcatel's announcements today for the 7450 Ethernet Service Switch, 7710 Service Router (SR), and 7750 SR lines expand that idea to create redundant relationships between boxes in a network.

"There have been people who maintain state across two processes in a box, but this is the first time I can recall anybody doing it across two boxes," says Mark Seery, an analyst with Ovum RHK Inc.

It's esoteric stuff, but high availability has helped Alcatel-Lucent get its routers -- created by startup TiMetra Networks, an Alcatel acquisition -- noticed by carriers. "It's something tangible they can test in the labs," says Lindsay Newell, director of marketing for AlcaLu's IP group.

That's led to a rapid rise in market share for Alcatel-Lucent, challenging Juniper for second place behind Cisco. AlcaLu has gotten a reputation for routing and Ethernet aggregation in IPTV networks, particularly. (See Routers, Switches Surge and Juniper Gives Static to IPTV Critics.)

Every router vendor has some interest in high availability, but a key for Alcatel-Lucent was to diverge from the approach favored by Cisco, called graceful restart. Instead, Alcatel-Lucent favored a method called stateful failover. "For many years the conventional wisdom was that it wouldn't scale, but so far it hasn't proved to be an insurmountable challenge," Seery says.

Alcatel-Lucent is releasing a handful of new high-availability features today, all of which are meant to work together to keep video streams, or any other traffic, delivered intact.

Take the case when the link goes down between an Ethernet aggregation box (like the 7450) and an access device (such as a DSLAM). Ethernet's standard means of finding a backup link takes too long for broadcast video purposes, so AlcaLu says its new approach is to allow the DSLAM to stay connected to two aggregation points, treating them as a single aggregation node.

That trick, known as Multichassis Link Aggregation, is coupled with a new, proprietary feature AlcaLu calls the Multichassis Sync Protocol, which lets the two aggregation boxes share information on each subscriber session in process. That way, in case of a failure, the backup connection can be activated immediately without losing information about any broadcast streams in progress. The whole idea is to minimize the time it takes for the network to recover from a failure.

Alcatel-Lucent routers can apply these features to any DSLAM supporting the IEEE 802.3ag standard for link aggregation. In other words, this isn't necessarily a bid to force carriers into all-AlcaLu networks. "For service providers who have other vendors' DSLAMs, this can drop in behind it," Newell says.

Alcatel-Lucent is also announcing high availability for virtual private LAN services (VPLS) and pseudowires. The pseudowire case is particularly important because redundancy has to be factored in for every new traffic type that hits the network. "Every time we have a new one, we have to deal with HA," Seery says.

Separately, Alcatel-Lucent has a few hardware announcements coming out today, including a six-slot version of the 7450 with 80-Gbit/s capacity.

— Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 3   >   >>
chechaco 12/5/2012 | 3:13:03 PM
re: AlcaLu Tunes Up Redundancy I'd like to point that it was not Alcatel/TriMetra who first delivered commecial version of stateful-based redundancy in routers but Avici. Avici went as far as modifying TCP/IP stack in order to provide hitless BGP session switchover. Not sure that Alcatel can do that.
Would appreciate any additional information regarding PW redundancy released by ALU.
vipul_rawat 12/5/2012 | 3:13:03 PM
re: AlcaLu Tunes Up Redundancy How different is it from Split Multi-Link Trunking (SMLT) that has been available on Nortel Ethernet switches for many years?

Some ref:
http://www.nortel.com/products...
http://www.nortel.com/products...
materialgirl 12/5/2012 | 3:13:02 PM
re: AlcaLu Tunes Up Redundancy For this to really work, how many extra boxes must one buy? For instance, if one link fails, and the neighboring boxes are operating at capacity, what is left to transmit the "extra" traffic? Can you fail over to just one neighboring box, or to a mesh?
sdmitriev 12/5/2012 | 3:13:02 PM
re: AlcaLu Tunes Up Redundancy PW redundancy in version 5.0 of TiMOS based on:
- MC-LAG (multi-chassis LAG) for access redundancy - it defines active/standby status
- Active/Standby PW between set ot PEs (access curcuit status signaled via T-LDP)
- Inter-Chassis Backup (ICB) spoke-sdp to provide fast recovery for "in-flight" packets (always active and used as last resort)
ip-eng 12/5/2012 | 3:13:01 PM
re: AlcaLu Tunes Up Redundancy Actually it was available on the Alcatel 7670 long before Avici announced it.
Pete Baldwin 12/5/2012 | 3:13:01 PM
re: AlcaLu Tunes Up Redundancy MG -- It doesn't sound like you can fail over to a mesh; I think there's just one redundant link involved. Checking with Alcatel on that.

Not sure you'd need any more extra boxes than usual. My impression (as an uninformed, non-network-building bystander) is that the new software takes your existing redundant network and makes failovers work faster. In other words, you'd need as many boxes as you were going to buy in the first place.

These are seat-of-the-pants guesses being made in the press room of the Game Developer Conference, so obviously don't take this as gospel. I'd welcome any other input...
chechaco 12/5/2012 | 3:13:01 PM
re: AlcaLu Tunes Up Redundancy As Tony Li once said: "I wrote bugs and will gladly write them again" (Hope I got it right). The point is - there's no such thing as bug free and 100% robust SW. Architectures and desings could be better or worse but to get to five 9s in Service Availability you must have NSF and redundancy in the system. Seems that ALU believes that redundancy in the network can substitute for redundancy in the box. Or is this technique complimentary?
Pete Baldwin 12/5/2012 | 3:13:01 PM
re: AlcaLu Tunes Up Redundancy I liked Mark Seery's answer when I asked him about Juniper's approach to high availability: "Building robust software in the first place."
paolo.franzoi 12/5/2012 | 3:13:00 PM
re: AlcaLu Tunes Up Redundancy "Alcatel-Lucent is releasing a handful of new high-availability features today, all of which are meant to work together to keep video streams, or any other traffic, delivered intact."

Care you note the words VIDEO STREAMS in this press release.

seven
tmc1 12/5/2012 | 3:13:00 PM
re: AlcaLu Tunes Up Redundancy brooks,

I know that you know everything but this is not just about IPTV. The MSE customers actually require some of these features more than IPTV. Pseudo-wire redundancy has nothing to do with Alcatel IPTV as they use VPLS for transport.
Page 1 / 3   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE