x
Optical/IP

Secretive Seranoa Scores $15.75M

Has stealthy Seranoa Networks Inc. cracked the code for IP savings? Or is it asking for a place in the poorhouse?

That's the multimillion-dollar question prompted by today's news that Seranoa has scored $15.75 million in new funding. Spokespeople say it's on track to make its product, an IP aggregation device, generally available by year's end (see Seranoa Scores $15.75M).

Seranoa claims to be "developing a new category of networking product, the IP edge concentrator," according to its public statement. The Seranoa Edge Concentrator, it says, sits in front of an edge router, grouping circuits together and replacing expensive WAN aggregation cards that would otherwise take up valuable router real estate.

Seranoa's not saying a lot more than this. "You can make up something, but we're not announcing product details," says VP of marketing Bob Olsen. But even though he won't say what interfaces the concentrator supports, one source familiar with the company says it's shunting IP circuits onto Gigabit Ethernet router ports, a further savings since LAN ports on high-end routers are cheaper than WAN ones.

Hang on: Is this really new?

The concept certainly isn't. "For a number of years, ISPs stuck frame relay switches in front of their routers just as aggregation devices, to save the cost of WAN ports," says David Passmore, research director at Burton Group.

Echoes of similar claims, albeit with different emphases, have been heard for years from makers of IP service switches, VPN gear, and edge routers (see A New Optical Taxonomy). Sadly, some of these are off the map, having added either too much or two little to the equation (see Sedona's Sad Demise).

Still, carriers like the pitch, and vendors soldier on, despite poor market conditions. CoSine Communications Inc. (Nasdaq: COSN), Nortel Networks Corp. (NYSE/Toronto: NT), and Quarry Technologies Inc. are in many ways trying to lighten the load of IP routers by handling services in front of them. So has Celox Networks (see Meet Me in Southborough, MA?), whose VP of marketing hails from the same company many of Seranoa's management do -- Net2Net, a network analyzer company sold in 1998 to Visual Networks Inc. (Nasdaq: VNWK).

Can Seranoa stake its claim in a market where so many players already are trying -- and in which some are failing?

There are hints that Seranoa Networks may have gotten an interesting new take on the IP aggregation formula:


  • It's based on network processors. Seranoa's working on network processors of its own design. That hints at cutting-edge smarts and scaleability, also being touted by key router makers that using these components (see Network Processors).
  • It's got traction. Seranoa has seven beta testers, a solid number, considering the lack of traction that's haunted some other vendors. Six of the testers are service providers, Olsen says, but one is an equipment vendor that may wish to peddle its wares in partnership with the startup. (A router vendor, perhaps?)
  • Startup savings claims hint at simplicity. Seranoa's claiming to cut per-subscriber hardware costs by three-quarters for ISPs, compared with what they're spending now for router interface cards. To do that, it will have to stay focused and simple, not trying to do too much.
  • Management team has experience. The management team has extensive backgrounds in ATM networking and management and analysis of LAN-to-WAN interfaces. This includes CEO Paul Kelley; VP of engineering and founder Phil Wilson; director of hardware engineering Stuart MacEachern; and director of software engineering Ralph Beck. Most of them hail from old datacom market leaders, such as Proteon, CrossComm, Primary Rate, Cabletron, and Racal-Datacom.

So Seranoa's got the good idea, the management, and now the money. What remains to be seen is whether it also has the ability to carve a niche where so many have broken the pick trying. The startup also faces formidable competition from edge router vendors such as Cisco Systems Inc. (Nasdaq: CSCO), which aren't likely to cede sales of interface cards easily. Still, depending on the identity of that beta tester, some interesting possibilities present themselves.

— Mary Jander, Senior Editor, Light Reading
www.lightreading.com
edgecore 12/4/2012 | 9:36:16 PM
re: Secretive Seranoa Scores $15.75M ----------

It's based on network processors. Seranoa's working on network processors of its own design. That hints at cutting-edge smarts and scaleability, also being touted by key router makers that using these components
-----------

Any idea which NPU they are using? An Intel IXP2400 comes to mind as its one of the few decent powered NPU's with dev platforms available to OEM's today?

EC


MrLight 12/4/2012 | 9:36:13 PM
re: Secretive Seranoa Scores $15.75M The 75% capex savings for internetworking between the edge of a service provider's network to the carrier's WAN network sounds compelling. The question is in what measure? Historically a 75% reduction meant a 4:1 concentration, meaning the box increases the fanout of the edge router by up to 4 times. Is that what seranoa means as well?

One point to take note of is these cost savings won't be linear. Also as you move through the WAN rates of DS-1/ADSL/DS-3/OC-3 etc and there are certain price break points i.e.a DS-3 will say cost only 4x to 5x a DS-1 per month, that will determine the actual savings. Meaning the 75% is the simplistic estimate based on router WAN port savings, the actual will be lower. However Seranoa is attacking a critical cost point for service providers, the cost to connect to the carrier's POP so I won't be too difficult on this point.

The other point is that OPEX is the big cost with edge routers, but it looks like Seranoa plans to have interoperability with IOS first which is both good and bad, since it ties them to one vendor until they get an JunOS interface.

Seranoa's feature claim is:
- "wide-area circuit aggregation" i.e from service provider edge router to the carrier's WAN
- "lower per-port costs"
- "multiplying service capacity" [of the edge router since more client ports can be connected and assuming the router is not WAN card throughput limited]
- "enabling automatic recovery from previously unrecoverable failures"

But they need more. For example SONET OC-3/SDH STM-1 channelized transport, SONET/SDH virtual concatenation or MPLS for circuit-switched multiple WAN destinations. It will be interesting to see if they plan to provide any of these once a product spec sheet is available.


A big concern I have that it isn't really clear how different Seranoa's product will be from a L2 switch with WAN ports, front-ending a router so the router can access links to multiple providers/destinations. If Seranoa's box was aLayer 3 (L3) switch no edge router is required as the L3 switch performs the routing function. [see giles post 334 in "MPLS will outgun ATM,says Poll" thread, where he talks about this as well]
so I assume it is a L2 switch.

So if Seranoa is just a better engineered L2 switch, Seranoa better think up some extra value to generate with all that NPU power at their disposal. They could use it to help with priority traffic management, MPLS QoS, router port load balancing, VPN acceleration for IPSec,etc.

MrLight :)
RoutedWorld 12/4/2012 | 9:36:07 PM
re: Secretive Seranoa Scores $15.75M
--------------------------------------------
Any idea which NPU they are using? An Intel IXP2400 comes to mind as its one of the few decent powered NPU's with dev platforms available to OEM's today?

EC
--------------------------------------------

Yea right. You should at least pick an NPU that's shipping. Or at least one that's been shipping for several months in order for Seranoa's boxes to be in trials. Oh, and do you mean power as in watts? Everything I read has the Intel NPU line as below average when it comes to performance.

My guess would be AMCC/MMC, Vitesse, or IBM.

edgecore 12/4/2012 | 9:35:59 PM
re: Secretive Seranoa Scores $15.75M ------------------------------------------
Yea right. You should at least pick an NPU that's shipping. Or at least one that's been shipping for several months in order for Seranoa's boxes to be in trials. Oh, and do you mean power as in watts? Everything I read has the Intel NPU line as below average when it comes to performance.

My guess would be AMCC/MMC, Vitesse, or IBM.

------------------------------------------------

RoutedDude,

Aside from your AMCC comment, Do you have any idea what you are saying?

Vitesse is getting out of the NPU business

http://www.lightreading.com/do...

IBM has an underpowered NP4GS3 and rumors have them looking at dropping out as well

Plus OEM's have been developing for the IXP2400 for a long time, using ADI Xscale boards and now moving onto the Angel Island. Is Seranoa's product fully developed yet, no, then why could their roadmap not have them in lock step with the IXP2400 rollout?

...EC

MrLight 12/4/2012 | 9:35:58 PM
re: Secretive Seranoa Scores $15.75M RoutedWorld.Many Choices, few products. I don't know the latest status of the Vitesse NPU but I thought it was dead as of Aug.02.

Check out www.e-insite.net/ednmag/index.... for Network processor info.

You know who has the BEST NPU right now, Cisco! Too bad they don't sell it on the merchant market.

I can't comment on who has the best in the commercial market since I haven't looked closely enough at all of them I have only looked at Intel, IBM, Agere, AMCC, and Zettacom.

Check ot the networking processing group form at www.npforum.org. The NP Forum assumed the activities of two previously existing groups, the Common Programming Interface Forum (CPIX) and the Common Switch Interface Consortium (CSIX) in 02/27/01. It has four working groups - hardware interfaces, software interfaces, performance benchmarking, and education.

The members are:
Actel
Agere Systems ( PayloadPlus NPU with fast cycle random access memory (FCRAM))
Alcatel
Altera
AMCC(nP3404 and nP7250)
Analog Devices
Avici Systems
Azanda Network Devices
Bay Microsystems
Broadcom
C-Port
Chiaro Networks
Chip Engines Ltd.
Cisco
Clearspeed
Conexant/Mindspeed
Corrent Corporation
Cypress Semiconductor
Dune Networks
Ericsson
Erlang Technologies
ETRI
EZ Chip
Fast-Chip (PolicyEdge network-classification processor)
Firebit Ltd.
Flextronics (Manufacturing)
Fujitsu Microelectronic America
Fulcrum Microsystems
Future Communications Software
Greenfield Networks
GSI Technology
HCL Technologies
Hifn
Huawei Technologies
IBM (PowerNP, Aug.02 $195 for NP2G )
IDT (IDT and Xelerated Packet Devices will co-develop line-card designs using Xelerated's X10NPU, also IDT has a 128k+ů72-bit full-ternary Internet Protocol coprocessor [email protected])
Infineon Technologies
Integrated Silicon Solution
Intel (IXP1200, IXP1250, IXP2400 priced at about $200, and 1.4-GHz IXP2800)
Internet Machines
IP Flex
IP Infusion (ZebOS Advanced Routing Suite SW)
Kawasaki LSI (100-MHz, 4.7-Mbit Content Addressable Memmory (CAM))
LSI Logic (algorithm-based CAM for ASICs)
LVL7 (Fastpath network-processing-development software for Intel )
Marvell International (Prestera-MX620 and MX630 10-Gbps NPUs)
Memcall
Micron Technologies
Modelware, Inc.
Mosaid (CAMs for single sycle access)
NEC
NetLogic
NetOctave
Next Hop Technologies
Nokia, Inc.
OKI Electric Industry
Paion Co., Ltd.
PetaSwitch
PMC Sierra (eliminated early summer of 2001)
Protean Devices
QQ Technology
RadiSys
Samsung Electronics
Sandburst Corporation
Sibercore (9-Mbit CAM)
Silicon & Software Systems
Silicon Access
Silicon Logic Engineering
SiPackets
SiSilk
Solidum (now iDT)
Sony Electronics
Spirent Communications (Test equipmenet)
STMicroelectronics
Sun Microsystems
Tau Networks
Teja Technologies (NPU softwaredevelopment platform, Teja NP, supports Linux)
TeraBlaze
TeraChip
TeraCross
Teradiant Networks (TN100 packet engine and TN101 traffic manager for 10-Gbit, TW200/201 equivalent for 20-Gbit, TN400/401 40-Gbit Note:Sept.02 In 10K the "packet engine family is price range at $1,250 for the 1,400-pin TN100 and extending to $2,600 for the 2,116-pin TN200 and 2,116-pin TN400. The traffic manager scales from $1,250 for the 1,100-pin TN101, to $2,600 for either the TN201 or TN401, both in 2,116-pin packages" )
TranSwitch
Tundra Semiconductor (Bridge ICs)
U4EA Group
Vitesse Semiconductor (NPU based on Colorado operations of SiTera and the North Carolina business of Orologic is on hold as of AUgust.02)
Xelerated (Working with IDT
Xilinx
Zagros Networks
Zettacom (ZSF Terabit Switch Fabric and the ZTM Advanced Traffic Manager product families)
Zte

The NP Forum introduced a IPv4 benchmark for comparing NPUs based IETF RFC2544, in August2002.

MrLight :) shedding some more light.
P.S.A good measure of what NPUs are really out there is look at how many are being supported by the embeded Linux vendors. I believe Seranoa is using Linux. So check out MontaVista,LynuxWorks, TimeSys, Consystant etc
RoutedWorld 12/4/2012 | 9:34:30 PM
re: Secretive Seranoa Scores $15.75M
------------------------------------------------
RoutedDude,

Aside from your AMCC comment, Do you have any idea what you are saying?

Vitesse is getting out of the NPU business

http://www.lightreading.com/do...

IBM has an underpowered NP4GS3 and rumors have them looking at dropping out as well

Plus OEM's have been developing for the IXP2400 for a long time, using ADI Xscale boards and now moving onto the Angel Island. Is Seranoa's product fully developed yet, no, then why could their roadmap not have them in lock step with the IXP2400 rollout?

...EC
--------------------------------------------------

Yes I do, and a little more then you.

IBM and Vitesse have stopped or pushed out development of 10G NPUs. Wow, in this market can you beleive that EC?

Vitesse didn't stop selling it's current products or did your Intel marketing group forget to tell you that? Same goes for IBM and C-Port who have also pushed roadmap products out.

So, let's do some simple math:

# of products shipping based on Intel 2400 = 0
# of products shipping based on Intel 12xx > 0

If you check you'll see that box groups are shipping products even today based on MMC, IBM, and Vitesse products.

Planning to use an IXP2400 and using one are two different things in my mind. It's hard to use something that doesn't exist. If Seranoa's planning to use an IXP2400 or any other Intel product that's fine with me.

Please go back to your Intel job and have a nice day.



RoutedWorld 12/4/2012 | 9:34:29 PM
re: Secretive Seranoa Scores $15.75M -------------------------------------------
MrLight :) shedding some more light.
P.S.A good measure of what NPUs are really out there is look at how many are being supported by the embeded Linux vendors. I believe Seranoa is using Linux. So check out MontaVista,LynuxWorks, TimeSys, Consystant etc
-------------------------------------------

Hello MrLight!

Interesting comment you made. Two years ago I would have said the same thing. However I'm really starting to see a number of new products giving up on embeded Linux and going with the Wind. WindRiver that is. Not saying I like it, matter of fact I don't, just a data point. They feel WindRiver is someone who can ride out the storm or just converging on a common OS (support more products with less people).

Cheers,

RoutedWorld (Linux fan)

MrLight 12/4/2012 | 9:28:13 PM
re: Secretive Seranoa Scores $15.75M RoutedWorld (Linux fan) "However I'm really starting to see a number of new products giving up on embeded Linux and going with the Wind. WindRiver that is. Not saying I like it, matter of fact I don't, just a data point. They feel WindRiver is someone who can ride out the storm or just converging on a common OS (support more products with less people)."

I like both.

I have worked on systems that utilized Windriver's VxWorks RTOS and its Tornado development environment and it did the job. I have also worked on systems using other embedded RTOSs and they likewise did the job. In the end I found it was the development environment that really is the differentiator. This will be the same for Linux.

I highlighted Linux only because I do not have access to WindRiver's or any other proprietary embedded RTOS vendor's NPU support plans.

It will be interesting to see which embedded RTOS wins the NPU sector; but it is a risky market for them with so many NPU platforms out there that they could align themselves with that may not make it. Only time will tell.

MrLight :)
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE