x
100G Ethernet

Non-Silicon Photonics Boost Arista's 100G

The on-board 100Gbit/s optics that Arista Networks Inc. recently announced are coming from Finisar Corp., according to one analyst. (See Arista's On-Board Optics Boost 100G Density.) The optics -- which include 10 lasers placed directly on a line card, replacing a transceiver module -- are making use of Finisar's Board Mount Optical Assembly (BOA), writes analyst Alex Henderson of Needham & Co., in a note published Thursday. Finisar has been shipping that product for more than a year. At OFC/NFOEC in March the company showed a version using 25Gbit/s VCSELs, which has the same footprint as the 10Gbit/s design. Arista's 7500E switch, announced last week, puts 12 100Gbit/s ports on each card, a feat achieved using Finisar's BOA, according to the analyst's note (though neither Henderson nor Light Reading has direct confirmation of this). That's a dramatic leap in 100Gbit/s density and one that comes at a critical time when viewed alongside silicon photonics developments, as the CFP2 interfaces that are just becoming available are likely to fit only eight ports per card. (See Silicon Photonics Prep for 100G Arrival.) Silicon photonics promises benefits in power consumption, cost and possibly density, but the few products on the market don't yet provide leaps like that. Henderson cites the 7500E as evidence that the silicon photonics threat to optical components vendors seems overstated. "Finisar is beating silicon photonics with dramatically lower-cost form factors and substantially improved capacity and enabling very low price-per-ports in the new Arista chassis," Henderson writes. Even so, Finisar isn't ignoring silicon photonics. "We already have access to the technology. We're looking at it extensively in our labs," Rafik Ward, Finisar's vice president of marketing, told Light Reading at OFC/NFOEC. "At ECOC [in September 2012], there were so many guys coming up to me with business cards. 'Hey, I'm from a fab.' 'Hey, I'm from a research organization.' There are people out there whose business model is to give silicon photonics access to anybody who needs it." — Craig Matsumoto, Managing Editor, Light Reading
dineshk913 5/8/2013 | 5:46:19 AM
re: Non-Silicon Photonics Boost Arista's 100G Would that avoid the cost of XFP or LC integration with that adds up the cost factor there. Though there may be port density but in the end its price and redundancy which customers look forward to ...
Lincoln Dale 5/8/2013 | 2:07:36 AM
re: Non-Silicon Photonics Boost Arista's 100G Some comments don't follow facts, not surprising for two anonymous posters.

1. "Have fun replacing the whole line card when a single VCSEL channel goes bad or dirty."
In the case of Arista 7500E-12CM-LC linecard, there isn't a lot of opportunity for dirt/dust to get in (automatic sealing flap on the MPO/MTP connector.) By a similar note, the MTBF is incredibly high, so failures aren't seen as a common thing even in very large deployments. Neither is an issue in the real world.

2."I don't think so, for 25G VCSEL, can't reach 150m which Arista has claimed."
Arista has not claimed its running at 25G @ 150m.
Arista has claimed 150m over OM4 MMF for STANDARD-BASED 100GBASE-SR10 which is 10x10 pMMF. And yes, not only has it been claimed, it can do it just fine.
xieyaohui 5/7/2013 | 11:55:50 PM
re: Non-Silicon Photonics Boost Arista's 100G I don't think so, for 25G VCSEL, can't reach 150m which Arista has claimed. Only 10G VCSEL at OM4 MMF ribbon.
reboot0 5/7/2013 | 5:00:58 PM
re: Non-Silicon Photonics Boost Arista's 100G Have fun replacing the whole line card when a single VCSEL channel goes bad or dirty.
reboot0 5/7/2013 | 4:57:48 PM
re: Non-Silicon Photonics Boost Arista's 100G Have fun replacing the whole line card when a single VCSEL goes bad or dirty.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE