x
Optical/IP

Cisco Acquisition Causes RPR Stink

Cisco Systems Inc.’s (Nasdaq: CSCO) latest acquisition has kicked up controversy at a technology standards body meeting this week in Portland, Oregon.

On Wednesday, the company announced the $150 million acquisition of AuroraNetics Inc. (see Cisco Buys 10-Gig Chip Maker), a 40-person semiconductor company that is developing a chipset for resilient packet rings (RPRs), a technology that allows data packet protocols like IP to work without sacrificing important resiliency and protection features of Sonet (see IEEE Tunes Ethernet for Telcos), a standard technology in most telecom networks.

Competitors expressed alarm that Cisco has scooped up a key supplier in the emerging market for RPR technology, because it may aid the data-networking giant in shaping the future of the standard. RPR technology is considered important because it will help enterprise data-networking technologies, such as Ethernet, be adopted in telecom networks.

“There are a lot of people expressing disappointment,” says Raj Sharma, director of product management for Luminous Networks Inc., a company also working on the RPR standard. “It makes it pretty hard to compromise with them now.”

The acquisition was announced in the middle of the Institute of Electrical and Electronics Engineers Inc. (IEEE) week-long meeting, at which representatives from different companies gathered in an effort to develop a standard for implementing RPR technology.

For over a year, Cisco has lobbied for its implementation of RPR called Spatial Reuse Protocol (SRP), a derivative of its early implementation called Dynamic Packet Transport (DPT), to be the basis for the standard. But companies involved in the working group charge that SRP has serious problems and is too limited in scope to be the foundation of the RPR standard. Cisco has been adamant about its own implementation and hasn’t publicly admitted that SRP has any problems.

AuroraNetics put an interesting twist on things. The semiconductor company has taken a gamble and started developing technology enhancements to SRP before any standards activity had even gotten into full swing. Two of its competitors already license Cisco’s existing SRP/DPT technology: Applied Micro Circuits Corp. (AMCC) (Nasdaq: AMCC) and Mindspeed Technologies. But others, such as Broadcom Corp. (Nasdaq: BRCM) and Vitesse Semiconductor Corp. (Nasdaq: VTSS), have taken a more conservative approach and seem to be waiting for the standard to get further along (see Cisco's Resilient Ring Gets a Boost and AMCC, Cisco Team on RPR Development).

The AuroraNetics enhancements to SRP had been a welcome change, says Sharma. For one, the company added a third class of service to the protocol, which originally had only two. This is an important improvement, given that IP QOS (quality of service) standards call for a minimum of eight priority classes. While adding only one more class hasn’t completely solved the problem, it's a start.

Cisco’s competitors in the working group hoped that AuroraNetics, an independent third party, could help push Cisco to improve its proposal and put it more in line with what others were working on. Now they worry that Cisco's purchase of the company will slow or stop that progress.

"If AuroraNetics had been left on their own, they would have evolved the technology and made it into a completely different animal,” says Sharma. "They were really the ones pushing Cisco to admit that SRP had problems."

In the past few months, Cisco had agreed to revise SRP. This was a big step for several reasons. For one, the changes to the proposal would make the new protocol incompatible with chipsets Cisco had already implemented in some of its products. Secondly, it was the first time that Cisco admitted SRP had problems. All of this was a good sign to smaller system companies that Cisco was willing to work and compromise for the good of the standard.

Now that Cisco will own the AuroraNetics intellectual property, the rest of the members of the RPR working group feel that Cisco will be less likely to compromise and work to improve the SRP proposal.

Some system vendors were also planning to use the chipsets developed by AuroraNetics, but, as one disgruntled CTO remarked at the meeting: “They are now are going to be forced to license the technology from Cisco."

Those attending the meeting this week are also baffled that Cisco would announce the deal in the middle of the conference. This puts Mike Takefman, a Cisco employee who is chairman of the working group, in a difficult situation.

But some people in the industry say that Cisco conspiracy theories may be overblown. David Newman, president of Network Test Inc., points out that Cisco has always been very active in the standards community and that its actions are often misinterpreted by its competitors.

"No vendor with the exception of IBM Corp. [NYSE: IBM], has been more of an advocate for open standards than Cisco," he says. "Have they tried to forward the Cisco way through standards? Absolutely, yes. But so has every one of its competitors." Cisco had not returned phone calls by press time.

- Marguerite Reardon, Senior Editor, Light Reading
http://www.lightreading.com

Larry, Monkey 12/4/2012 | 8:05:54 PM
re: Cisco Acquisition Causes RPR Stink Thanks, Lopez.
Duly corrected.

LR Copy Chief
Lopez 12/4/2012 | 8:05:54 PM
re: Cisco Acquisition Causes RPR Stink It is Mike Takefman, not Mike Takesman, please correct. You can delete this message once you've corrected it too.
metroshark 12/4/2012 | 8:05:51 PM
re: Cisco Acquisition Causes RPR Stink It is true that the timing of Auroranetics acquisition was really bad. However, the comments in the article about "SRP having problems" is not fair. SRP is so far the only resilient packet ring protocol presented to 802.17. Other companies talked about some general high level aspects of some alternative solutions, but not one of them has presented a single complete and detailed proposal so far. It is pretty easy to find faults when the information is out in the open, and almost all standards (including Ethernet) has shortcomings and certain corner cases. The goal of a standards effort is not to find the best possible solution in the world, but to come up with a solution that works and provides a common basis for interoperability. I rather go with SRP as the basis for 802.17 RPR standard at this point, as opposed to risking one of the other proprietary implementations which are neither documented nor implemented.
prefer_to_lurk 12/4/2012 | 8:05:49 PM
re: Cisco Acquisition Causes RPR Stink "SRP is so far the only resilient packet ring protocol presented to 802.17."

I don't think that's true...

Nortel has OPTera Packet Edge -- which is actually working, very well in fact, and doesn't rely on Routers at the edge (as does DPT).

Luminous is also actively pitching something, and I think Dynarc is too.

Cisco is actually somewhat BEHIND in this area. Maybe that's why they made the acquisition ?

ptl
techmedia 12/4/2012 | 8:05:38 PM
re: Cisco Acquisition Causes RPR Stink Is it really likely that Cisco could take control of the RPR standards process? What would be the impact on companies like Dynarc who have already made a couple of deployments?
prefer_to_lurk 12/4/2012 | 8:05:36 PM
re: Cisco Acquisition Causes RPR Stink
Silicon for Nortel's OPE implementation of RPR is also available to other vendors:

http://biz.yahoo.com/bw/010423...

ptl

metroshark 12/4/2012 | 8:05:36 PM
re: Cisco Acquisition Causes RPR Stink It is true that there are other proprietary resilient packet ring implementations out there, however, Cisco's SRP is the only one for which all implementation details have been publicly disclosed. Also, silicon for 2.5 Gb/s version of SRP is available from multiple silicon vendors today.

Some of the other system vendors have promised numerous times that they were going to present a proposal based on their RPR implementations but so far, nothing other than a few power point slides have been presented to 802.17. It is pretty hard to evaluate the pros and cons of alternative technologies as long as these vendors keep their implementation secret.
fundamental_guy 12/4/2012 | 8:05:17 PM
re: Cisco Acquisition Causes RPR Stink Cisco does have a proposal called SRP-fa.

Nearly 3 or 4 years ago Cisco documented this proposal
in an RFC and went to IETF to get this endorsed.
IETF is Cisco's backyard and they know how to
get things done in IETF. If you look at the
"old boys club" that Cisco has in place at IETF
you would be amazed to note that Cisco failed
miserably in IETF to get SRP-fa endorsed. That
fateful meeting was a disaster for Cisco. They
were asked to take this proposal to IEEE since it
dealt with layer 2 protocol and was not really
a layer 3 (IP) protocol.

Cisco has very little or no experience in IEEE
standards. In IETF two vendors can describe an
implementation, talk about interoperability and
get a end user to endorse it to create a standard.
IEEE is a completly different process. I think
Cisco is learning this the hard way. The more they
lobby for SRP-fa to be 802.17 baseline the more
resistance they will have and the damage to their
DPT/SRP technology and installed base.

The good part is that at the last IEEE meeting in
Oregon Cisco finally admitted in the open that
SRP was NOT designed for the diverse market
application that is currently seen by the industry at large.
Specifically, Current DPT/SRP products do not
support (and cannot without major modifications
to SRP) the following:
1. Ethernet private line services.
2. Transparent LAN services
3. Circuit emulated TDM services.

DPT/SRP was primarily designed to eliminate the ADM used
in SONET networks to support aggregation and
de-aggregation on internet traffic between
access router and backbone routers using POS links.
This focus was key in making DPT/SRP happen
quickly for Cisco but now has become the limiting
factor of DPT/SRP being accepted as a baseline
for the standard.

There are many vendors in RPR who are shipping products. Obviously, these are documented
somewhere. Disclosure of these will only happen
when Cisco openly admits that SRP may not make it
as the ultimate standard. Since this has happened
now we should see the others put together a
proposal.

It is always easy to get a single a company
to define a technology and standardize it.
I wonder if the industry would see the creativity
it has seen so far with this approach.
It is always a difficult job to get competing
vendors to compromise on a single standard.
However, the later brings a lot of diversity
in thinking that is required as a part of the
due dilligence in standards and it also forces
convergence without companies loosing their face.

It is unfortunate that Cisco still does not get
it. Times have changed. Cisco now must address what
is perhaps the most difficult business issue moving forward.
- How will current DPT/SRP users be upgraded
when the new RPR standard comes in?

If Cisco does not pay attention to this they will
not only loose the current battle in IEEE but will
loose the war on multi-node shared rings.

riteleading 12/4/2012 | 8:03:12 PM
re: Cisco Acquisition Causes RPR Stink I would be interested to understand how SRP cannot provide:
1. Ethernet private line services.
2. Transparent LAN services
3. Circuit emulated TDM services.

It is clear that Cisco does not provide the above
on a router. But that is the problem with buying
a router, if you want the above services Cisco is
not your box vendor. That being said....

Based on my reading of the RFC, I am not sure that
these are services that cannot be provided by SRP.
SRP is a basic MAC layer and provides the
basics for doing all of these things. A low
jitter / delay path that can be used for implementing TDM emulation. Encapsulation of an Ethernet packet within an SRP packet will allow both 1) and 2) to be implemented.

It will be interesting to see what these other companies do come up with and how it does compare to SRP. If some of the ideas that have been presented by some of the companies make it through then there will be a clear choice. A packet ring that looks and feels somewhat like other 802 standards, or some kind of ATM-like hodge-podge that people might vote in to punish Cisco for being a bit ahead of the curve and a little too brash about it. Not that kicking the gorilla
isn't fun!

802.17 could do a lot worse than SRP, and it is
not yet clear that it will do better.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE