Comms chips

Next Gen Sonet: What About SDH?

Plenty of vendors have brought out next-generation Sonet chips in the past few months, paving the way for data traffic to be carried more efficiently over existing carrier infrastructure.

Some of these chips also claim to support SDH, the transport protocol used in telecom networks everywhere outside North America. But everyone's dirty little secret is that, in fact, they can't support next-generation SDH as effectively as they can support next-generation Sonet.

That is, until yesterday, when startup West Bay Semiconductor Inc. announced its WB4500 virtual concatenation framer (see West Bay Ships Sonet Chip). The startup claims it has the only chip that brings the full benefit of the virtual concatenation technique -- a key element of next-gen Sonet -- to SDH.

Why? There are two issues: encapsulation protocols and granularity. Time to dig into the details.


For starters, there's the issue of the best way to encapsulate packets for transportation over an SDH network. The Yanks have pretty much decided to do this using a protocol called generic framing protocol or GFP. GFP supports both Sonet (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy), in theory.

Unfortunately, life isn't so simple outside America. Other countries are insisting on support of an alternative called link access protocol–SDH or LAPS (see Next-Gen Sonet ).

LAPS is a required standard in China -- a huge potential market for SDH equipment -- so vendors that don't support it won't get very far selling out there, says West Bay's president and CEO Tino Varelas. LAPS is also on the rise in Europe. Even if it's not a requirement, supporting the "local" standard helps network equipment manufacturers feel more at ease, he adds.

West Bay's WB4500 supports LAPS, as do the HDMP3002 from Agilent Technologies Inc. (NYSE: A) and the EtherMap48 from TranSwitch Corp. (Nasdaq: TXCC). Both the Agilent and TranSwitch chips were also announced yesterday (see Agilent Intros Sonet Chip and TranSwitch Expands EOS Line).

However, support of LAPS is not enough, says Varelas. The other issue is:


In the Sonet world, virtual concatenation involves glomming together the smallest channels -- STS1s (51.8 Mbit/s) -- to make suitably sized pipes for the particular service to be carried. There's no direct equivalent in the SDH world, where the smallest channels, called AU4s, are three times the size –- 155 Mbit/s. (For more on this issue, see: Tutorial on Grooming Switches.)

As a result, vitual concatenation provides a less efficient solution for SDH than it does for Sonet. For example, a Fast Ethernet (100 Mbit/s) channel can be carried by two Sonet STS1s, thus occupying nearly 100 percent of the bandwidth. In SDH, however, putting a Fast Ethernet channel into 155 Mbit/s results in only 67 percent bandwidth utilization.

West Bay says it overcomes this problem by building a chip that can see down to the VC3 level. The VC3 is the same size as an STS1 but is hidden by several layers of protocol origami. Each VC3 must be mapped into what's called a TUG3, and then three of them are combined to form a unit called AU4. West Bay's chip can deal with these complexities without worrying about the bizarre math that adds up three and gets four.

The upshot is that West Bay claims to offer the only sensible solution for virtual concatenation outside America -- although Transwitch is also claiming virtual concatenation at the VC3 level in its product announcement today.


Inside America, West Bay's closest and most serious competitor is probably Agere Systems (NYSE: AGR). Agere is the incumbent in virtual concatenation chips and is the only company shipping product right now (see Agere Produces Sonet Framer Chip).

Many of the other wannabe vendors in the virtual concatenation space -- namely Agilent, Cypress Semiconductor Corp. (NYSE: CY), and PMC-Sierra Inc. (Nasdaq: PMCS) -- have chosen to optimize their chips around the specific application of sending Gigabit Ethernet over Sonet. PMC-Sierra and Agilent, for example, have integrated the Ethernet MAC (media access controller) with their framers, which has the advantage of allowing them to interface directly with optics.

West Bay and Agere, on the other hand, have opted for something more flexible. They are targeting a wider range of applications such as aggregation and multiservice provision platforms. Both include an STS1-level crossconnect that can groom TDM (time-division multiplexed) traffic before sending it off across the network.

A distinguishing factor from the competition, says West Bay, is that its chip supports up to 48 independent logical channels, which can each carry different protocols -- including Asynchronous Transfer Mode (ATM), packet over Sonet, GFP, or LAPs -- simultaneously.

"For many applications, the only chip that will work is the West Bay chip," boasts Varelas. If a customer wants to aggregate Fast Ethernet channels, for example, then he or she will need the West Bay chip because it is the only one that supports enough logical channels, he says.

For comparison, Agilent and TranSwitch say their chips support four channels. Agere did not respond to calls.

West Bay expects samples of its chip to be ready in the third quarter. The WB4500 is priced at $475 in quantities of 1,000.

— Pauline Rigby, Senior Editor, Light Reading
Page 1 / 2   >   >>
rafaelg 12/4/2012 | 10:19:56 PM
re: Next Gen Sonet: What About SDH? "...involves glomming together..."

Perhaps gromming?

Opti-Mystic 12/4/2012 | 10:19:41 PM
re: Next Gen Sonet: What About SDH? Regarding: "Three VC3s must be wrapped up together to form what's called a TUG3, which is then mapped into a unit called AU4."

Not quite. A VC3 is mapped to a single TU-3, which is then mapped into a single TUG-3. Then a total of (3) TUG-3's are wrapped together to form a VC4, which then is mapped to an AU-4. The bottom line is that there are 3 TUG-3's in an AU-4, not one, as the article stated.

- The OptiMyst
lightpack 12/4/2012 | 10:19:38 PM
re: Next Gen Sonet: What About SDH? You are right! Moreover Cypress's chip also supports TUG 3 mapping and both PMC and Cypress's chips support Virtual Concatenation. They need to get their facts right....
lightpack 12/4/2012 | 10:19:38 PM
re: Next Gen Sonet: What About SDH? Hmmm - LAPS is essentially X.86 and purely for Ethernet over SDH. Moreover it uses similar delineation as HDLC/PPP.
GFP is there for multi protocol and is for both SONET and SDH. Don't see how they compete - if anything GFP can do what LAPS does i.e. Ethernet over SONET and more! You have Framed GFP for PDU oriented adaptation mode and Transparent GFP for block-code oriented adaptatiion mode.
wass 12/4/2012 | 10:19:35 PM
re: Next Gen Sonet: What About SDH? Which PMC chip supports Virtual Concatenation and TUG 3/VC-3 mapped data today (GFP and LAPS) (and at OC-48/STM-16)? That's the whole issue. There are very few that do. Cypress is one additional company that does.

I have to laugh at the way the article makes it sound like West Bay invented VC-3 VC. VC-3-Xv is a basic part of the SDH standard for virtual concatenation.

I noticed no discussion about LCAS in the article as well. This seems to be a shortcoming in all of these devises.
bad boy 12/4/2012 | 10:19:26 PM
re: Next Gen Sonet: What About SDH? Pauline,

I wonder if you are also getting Cypress and PMC' comments on the tall claims by Westbay!

As far as the sampling of real chip with Virtual Concat goes, these are the first two companies who have sampled devices in the market.

Pauline Rigby 12/4/2012 | 10:19:25 PM
re: Next Gen Sonet: What About SDH? Well, I talk to PMC-Sierra and Cypress about their products, and covered both their virtual concatenation stuff when it came out. I haven't talked to them specifically about West Bay -- there's only so much time in the day.

I've heard rumors that PMC's chip isn't ready yet -- there are some bugs that need fixing -- so it isn't actually shipping. It will probably ship in a couple of months. I did hear this from a competitor, but the competitor then when on to say that their chip had bugs too and wasn't ready either!

Which part of West Bay's claims do you find hard to believe?

[email protected]
bad boy 12/4/2012 | 10:19:24 PM
re: Next Gen Sonet: What About SDH? "Which part of West Bay's claims do you find hard to believe?"


I am talking of their claim for "sensible solution for virtual concatenation outside America"

Here are my comments - both on West Bay's claims and the article:

- West Bay is not the only one who knows VC-3 and TUG-3 etc. West bay is not the first one to support this.
SDH is pretty old and people like Agere and PMC surely know this. Even new comers like Cypress know this and have support for TUG-3 mapping in their VC part.

- LAPS is a variant of POS (HDLC-PPP): Initially it was only header variation, now it seems there is little more to it. BUT not so much different. Most of these chip guys supporting HDLC delineation can support LAPS hands down.

- Agilent and PMC chip has integrated GbE MAC - This is the largest section of the market which needs VC - and they provide most integrated solution (It is a different matter that none of these two are proven to be working - neither is West bay for that matter!)

- Agere and Cypress have most flexible part - and now West Bay can be added to this category - which can work for multiple protocol (GbE / FC / ESCON etc.) at the cost of less than optimum integrattion for a given protocol.

Similar to your PMC comment, I have heard Agere part not having Vitual Concat function working right. If these are true, all said, Cypress is the only Virtual Concat part working in the market today (untill you hear something else for Cypress part as well!!!)

Have fun!!

aardvark 12/4/2012 | 10:19:08 PM
re: Next Gen Sonet: What About SDH? Bad Boy,


I've checked out all the material available and I can't see any reference to the Cypress device supporting TUG-3.

As far as I know the only device available with TUG-3 is PMC's SPECTRA-2.4 and for some bizare reason they have dropped this function for the SPECTRA-10G.

If the Cypress part does support TUG-3 then that's great but until I see it in writing I can only assume that it doesn't.

Can you point me to the smoking gun.

asbjorn 12/4/2012 | 10:19:00 PM
re: Next Gen Sonet: What About SDH? Is there anything LAPS can do that not is support of GFP?
Page 1 / 2   >   >>
Sign In