Thanks Odo. It's true Intune got a good chunk of money, but Matisse has been at this longer ... and we all know stories about startups that didn't succeed despite a stash of funding, right?
Kinder and gentler? Well, maybe we're just in award-giving mode. That doesn't mean we won't bash on people later. :)
Abashford -- What we like about Cyan is the concept more than anything else: a chameleon-like packet-optical god box. In picking something that represents the direction of the industry in 2010, I'd say it's a strong choice.
What do you find so impressive about the Cyan solution??
The Z33 is a nice dense MDWDM platform (whoopie..)
The Z77 moved the optical patch panel from the front to the back (great, as long as you have access to the back of the shelf). Nice converged platform for those with the luxury of converged operations. I suspect with all those electrical and optical pins on the backplane, it is bound to be expensive.
Flashy GUI... again, cool if you are doing green-field.
This one was difficult. We decided we liked Optical Burst Switching, although it's a bit controversial on these boards. But Cyan edged out Matisse in the North America category.
Had it come down to Matisse vs. Intune, I think Matisse might have won out, for being further along commercially (as far as we can tell). So -- apologies to Matisse, a very close runner-up here.
AlcaLu's converged transport strategy was also a strong contender and certainly something to watch for in 2010.
Cyan Optics Inc.'s Z77 and Z33 multilayer cross-connects
Cyan's modular systems represent a new kind of packet-optical transport system that can operate at multiple layers in the network.Others are likely to come up with similar concepts -- Ciena's 5400 fits that mold -- but Cyan is out of the gate early, and it has the added punch of being a rare optical startup with customers in deployment.
Keeping in mind that Top Picks is forward-looking, we're giving the nod to the promise of OBS. Intune and Matisse -- a close runner-up in the North America category -- are developing optical ring schemes that could bring new packet-optical capabilities to bear in the metro network.
As pointed out on these boards, there's no guarantee it will work -- but it's certainly a technology to watch. Intune hasn't launched products but does have a contract with the Irish government; we'll see just how serious that is.
As carriers begin to revamp the optical transport network, the 8800 seems to be pushing all the right buttons: OTN transport with grooming capabilities, ROADM functionality, and a claimed 1.28 Tbit/s switching capacity.
Interesting note from Xilinx, which wants to be considered for the Virtex-6 HXT FPGAs, which are 40/100-Gig ready. Here's their pitch:
* * * * *
1.Integration – Bandwidths increase, but form factors are not growing with the increased bandwidth. So designers have to do more with less (area). Integration of tons of logic with the 10G & 6.5G transceivers enables designers to achieve this;
2.Bandwidth – HXT integrates 72 serial transceivers in a programmable device for the first time in history. This provides designers with bandwidth options that were not previously available without paying for it dearly in area and power dissipation. Integration of 10G transceivers saves up to 80% of transceiver power compared to an external Serdes option;
3.Package Signal Integrity - Virtex-6 incorporates the most optimal pinout and package design to provide system designers with superior signal integrity. The package is designed to isolate parallel IO noise from the serial IO, and over 30 dB noise isolation between channels and over 40 dB isolation between Tx and Rx;
4.Targeted applications, from Optical Transport Network (at 100G OTU4) to Data Center Access (100G Ethernet), from High end packet processing capabilities (100G Traffic management) to high efficient Low end Access systems policing (10G Traffic manager in a low end Spartan 6) are really changing the way Xilinx FPGAs drive next generation networks.
Regarding ALU’s converged backbone: - There may be incremental benefits for the reasons mentioned (though when considering the full cost and risks of moving to a more complex integrated network, at least some of the stated integration benefits may be overturned), the actual need appears to be for quite radical cost efficiency increase: If with existing, segregated service (IP) and transport layer networks the economics for network services are marginal at best, as the rapidly diverging revenue/bit and cost/bit trends continue, it appears that more radical changes in the technology and service architecture and business models are needed to make a sustainable business case for carrying the bulk of the traffic on the ‘net today and going forward. - Secondly, due to the unavoidable complexity of implementing and supporting networking platforms integrating functionality from L0 optics through L1 muxing (what eg ODU really is, using plain TDM) and L2 switching, VPNs to L3 IP routing and beyond, the SPs will be left with a very limited number of capable suppliers for such converged network technologies, causing for SPs the same problems of vendor pricing control if they took the fully converged network platform approach as they currently have with the very limited number of alternative vendors for IP routing gear. - Thus, as opposed to mere multilayer integration of technologies, it appears that more transformative approaches should be considered, eg: *1 Look for optimization beneath a client layer boundary that allows, where desired eg for operational or contractual reasons, to keep the horizontal layers independent of each others, while delivering cost efficiency in terms of service produced at given layer. Ideally, one should figure out how to improve the revenue/cost performance of a layer (n) service while allowing to keep layers (n+1) as is, and then the net benefits would be quite clear. *2 Instead of arguing about why given technology direction is best, when will we begin to see more hardware-as-service type business models (e.g. from startups, as we’ve seen in the software world?) where the customer only buys the revenue-cost benefits of the given network function without concerning oneself with how that function was implemented, as long as it transparently serves the client layers? Where's the salesforce.com of networking gear?
And to Craig - OBS has obvious promise, but as mentioned, it appears that the real concerns are the operational aspects, incl SPs maxim of fighting additional complexity at all costs. Ideally it should operationally be plug ’n play & leave alone, like a standard static mux, while radically improving the revenue-cost difference per bit (see *2 above). Is OBS, or any other contestant there yet? Will they take us there?? Or will the era of relatively abundant and inexpensive (core) network bandwidth, and the growingly frivolous net applications like u-tube, FB etc, be behind us after one or two telco’s standard capital investment cycles when the investment was not recovered??? Will the network again become predominantly a boring conduit of serious business communications... or will the users agree to pay for internet apps to cover their true cost over time..?
The blogs and comments are the opinions only of the writers and do not reflect the views of Light Reading. They are no substitute for your own research and should not be relied upon for trading or any other purpose.
To save this item to your list of favorite Light Reading content so you can find it later in your Profile page, click the "Save It" button next to the item.
