re: 100-Gig Ethernet Gets Official NodSome way to go still !! Access still would be circuit switched using Pseudowire packet core ... how many operators want to overhaul their network in a go ?? Anyone on the fiber requirement for transpoprting 100G ?? distance support ??
re: 100-Gig Ethernet Gets Official NodPacket Over SONET is not dead yet either. POS interfaces on core backbone and edge routers in ISPs and Carriers are still shipped like in GSR and Juniper's routers. So, Carrier Ethernet replacing SONET as the backbone ring transport mechanism is only one of the replacements. But, Metro Routers seem to be built these days only with Ethernet transport and services in mind.
re: 100-Gig Ethernet Gets Official NodEthernet sure has taken a big jump from 10 to 100G, whereas SONET still at 40G.
In my opinion might be some other reason for not carrying over a single wavelenth, is it flexibility as ethernet directly over Fiber is not dynamic due to limitations of Optical ADM,crossconnect etc. I know both are avaialble but not as good as SONET ADM & Crossconnect.
re: 100-Gig Ethernet Gets Official NodWill be IP capable of scale to these speeds?
We need a new generation of internetworking equipment faster, without routing or bridging tables, learnings, label swaping or storage and forward mechanisms.
re: 100-Gig Ethernet Gets Official Nod10 Gigabit Ethernet is not hitting the right price points yet for widespread & main stream (Enterprise) deployments - can you imagine the price of a 100 Gigabit Ethernet port (and all the associated system upgrades it would take to deploy such a product)?
I think that 100Gbps Ethernet will be first deployed in service provider environments - it won't be enterprise environments that drive the price of the technology down, unlike 10/100/1000 Mbps Ethernet. 10GigE is probably a half half situation - the downward pressure on price is coming or will come from both enterprise and service provider markets.
As for SONET/SDH verses 100Gbps Ethernet ? It will all come down to what you need and what you can afford. If you "need" SONET/SDH but can't afford it, you'll probably change your needs to suit what you can afford i.e. the needs that are compromised weren't really needs in the first place if they can't be afforded. If they are true needs then you should be able to get more money to by the technology you need.
One thing I also think effects the SDH/SONET vs Ethernet comparison is the types of applications that are being run over networks these days. Most applications are now or very rapidly moving towards packet based, where occasional packet loss is expected and can be dealt with. I don't know that much about SDH/SONET however I understand one of its fundamental design requirements was the ability to carry TDM style traffic, which is where requirements such as 50ms switch over etc. probably came from. Packetised traffic Over Sonet is obviously something that works, however the TDM oriented features are an expensive overhead when the traffic carried doesn't require them. Ethernet doesn't have those feature "overheads", and therefore those feature overhead costs, which is one of the things that makes it cheaper when providing the same level of bandwidth as an SDH/SONET link.
re: 100-Gig Ethernet Gets Official NodThe "primary" selling point of high-speed Ethernet be it Gigabit, 10 Gigabit, 100 Gigabit, 1000 Gigabit, 1,000,000 Gigabit, etc...is the feature/cost benefit(s).
The long-range promise (I'll still love you in the morning...) of this technology is the potential lower total cost (in term of components and installation/maintenance) HOWEVER 10 Gigabit Ethernet has YET to prove that it can hit the component prices touted/forecasted - one single 10 Gigabit Ethernet port (in terms of a total system cost) is still more than 10x the price of ten single Gigabit Ethernet ports trunked together.
100-Gig is a 'nice to think about' road to nerd-vana but the reality of widespread adoption is not going to appear for 10 years or so (and this is a skeptical predication at that).
10 Gigabit Ethernet is not hitting the right price points yet for widespread & main stream (Enterprise) deployments - can you imagine the price of a 100 Gigabit Ethernet port (and all the associated system upgrades it would take to deploy such a product)?
re: 100-Gig Ethernet Gets Official NodIt has always amazed me how many people think that TDM SONET is complicated. I remember the frustration of trying to teach TDM to packet people during the optical metro networking boom/bust cycle, and how many just couldn't comprehend that all SONET was trying to do was to be a wire.
That continues to be the thing that many people miss: Ethernet makes a great packet transport mechanism, but it makes a crummy wire.
re: 100-Gig Ethernet Gets Official NodMr Zippy wrote: ..."Packetised traffic Over Sonet is obviously something that works, however the TDM oriented features are an expensive overhead when the traffic carried doesn't require them. Ethernet doesn't have those feature "overheads", and therefore those feature overhead costs, which is one of the things that makes it cheaper when providing the same level of bandwidth as an SDH/SONET link."
From a complexity point of view, SONET TDM devices are about as simple as they come, much simplier than packet switches. But SONET devices don't have the mega volumes of end customer centric Ethernet devices, so they cost more. SONET has the necessary payload channel, embedded operations channels and operations systems for high volume, high bandwidth networks. The package of features has been worth a premium but of course there is continuing price pressure.
Real-time interactive data needs reliable low drop rates and low latency characteristics. High volume Ethernet switches are cheaper than SONET but cheaper does not mean its extensible to continental distance networks. For some key applications, adding many packet switches along a network chain is equivalent to adding springs in a drive chain on a car. VOIP users and interactive gamers will demand good latency performance: for this type of user a good channel is more than just a bandwidth number. Service providers spend a great deal of time designing their networks to assure services do not pass through too many packet switches. As the performance of packet switches and their management systems improve, they will continue to grow into more places in service provider networks.
re: 100-Gig Ethernet Gets Official Nod"one single 10 Gigabit Ethernet port (in terms of a total system cost) is still more than 10x the price of ten single Gigabit Ethernet ports trunked together."
I'm not sure that's the whole story (even if it's true).
the fact is that:
1) it's harder to manage a bundle of 10 GE links than a single 10GE link.
2) load balancing of large flows (where "large" is anything that's a significant fraction of the individual link speed) over multiple links is tricky.
3) you need to include transport cost in the equation (even if you're just talking about a bunch of fibres across the building).
So I guess with respect to my third point the big question comes down to availabilty of serial transmission solutions for 100GigE that can sit in a WDM channel - though granted burning a fibre for 100GE is probably OK at least intra-metro.
Anyone on the fiber requirement for transpoprting 100G ?? distance support ??