The SPB Versus TRILL Debate

Michael Vizard

There's a debate in network circles these days about the future of Ethernet switches that is roughly equivalent to the argument between VHS and Betamax that took place when video cassette recorders first came to market.

On the one side is a group of networking vendors that support an emerging IEEE 802.1aq Shortest Path Bridging (SPB) approach, while another group supports an equally emerging Transparent Interconnection of Lots of Links (TRILL) approach that is backed by Cisco, Juniper Networks and Brocade.

Most recently, the backers of SPB, which include Alcatel-Lucent, Avaya, Huawei, Solana Networks and Spirent Communications, got together to demonstrate interoperability between next-generation networking gear that supports SPB. According to Roger Lapuh, a product line manager for Avaya, SPB represents a more natural extension of existing switch technology in contrast to TRILL-based network gear that would change the way networks are fundamentally deployed and managed.

Both SPB and TRILL are trying to address the need to replace the archaic spanning tree protocol. Both approaches make it possible to put more intelligence in the network layer. That's critical because we need the ability to track virtual machines and application workloads across the network. Once we have that capability, we can then more easily manage applications and their associated IT infrastructure as a single logical entity.

It's unclear which of these standards will ultimately prevail. It may be that both approaches will eventually be adopted under some common architecture. Whatever happens, there needs to be a resolution of the debate sooner than later because as each new virtual machine proliferates across the network, our ability to effectively manage the network deteriorates that much more, which at the rate we're on, should become a full-blown crisis sometime next year.
 



Add Comment      Leave a comment on this blog post
Sep 22, 2011 7:09 PM Markus Nispel Markus Nispel  says:
As part of our unified fabric architecture we at Enterasys will also support IEEE 802.1aq SPB and we recommend this standard to our customers. From our point of view it provides a much more seamless migration from existing RSTP/MSTP and virtual switch solutions into a standards based meshed infrastructure. The support of other IEEE standards like DCB Data Center Bridging, Ethernet OAM and others makes SPB more attractrive. In terms of unification of the infrastructure for mid size customers SPB can also be used in a campus infrastructure and so customers won�t expirience any technology break. As the IEEE did standardize layer 2 topology protocols for the last decades we expect this the case for the future as well. We plan to participate in the interop tests mentioned before as well in the future. Reply
Oct 5, 2011 7:10 PM Peter Ashwood-Smith Peter Ashwood-Smith  says:
Couple of comments. 1) I'm not aware of any public support by Juniper for any standards based L2 fabric work, their Q-fabric is neither TRILL nor SPB. 2) Markus , we look forward to you bringing an Enterasys switch to our next SPB Interop. 3) I don't think having several choices for L2 fabric control is a "full blown crisis". On the contrary, competition between different standards and proprietary solutions ultimately drives innovation up and costs down. After all if you don't like competition .. just stick with STP ;) The VHS v.s. Beta is not really a good analogy because your device could only play one type of tape and movies had to be mass produced to one type. The new crop of switches with ASICs or NPUs can likely support either or both of these protocols with just a new software load, although older switches of course can't run the new TRILL data paths. Reply
Oct 15, 2011 11:10 PM Jon Hudson Jon Hudson  says:
Peter is dead on with several points, and I'll go further.... 1.) Juniper has actually been very very ant-TRILL, certainly not ever for it 2.) If anything Juniper has come out in support of SPB, but QFabric is neither TRILL or SPB 3.) Huawei supports both. One of the co-chairs of TRILL is a Principal Engineer at Huawei. 4.) BETA vs VHS is a very poor analogy. OSPF & IS-IS is much more acurrate a comparessen. 5.) This isn't Highlander, one must not die for the other to survive. 6.) In a few years most vendors will support both. 7.) http://ethernetfabric.com/2011/10/why-can�t-we-all-just-get-along/ Reply

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.