Rebranded Ethernet Technology Consortium Unveils 800 Gigabit Ethernet (anandtech.com) 29
The Ethernet Technology Consortium, the non-IEEE, tech industry-backed consortium formerly known as the 25 Gigabit Ethernet Consortium, has announced a new 800 Gigabit Ethernet technology. AnandTech reports: As for their new 800 Gigabit Ethernet standard, at a high level 800GbE can be thought of as essentially a wider version of 400GbE. The standard is primarily based around using existing 106.25G lanes, which were pioneered for 400GbE, but doubling the number of total lanes from 4 to 8. And while this is a conceptually simple change, there is a significant amount of work involved in bonding together additional lanes in this fashion, which is what the new 800GbE standard has to sort out.
Diving in, the new 800GBASE-R specification defines a new Media Access Control (MAC) and a Physical Coding Sublayer (PCS), which in turn is built on top of two 400 GbE 2xClause PCS's to create a single MAC which operates at a combined 800 Gb/s. Each 400 GbE PCS uses 4 x 106.25 GbE lanes, which when doubled brings the total to eight lanes, which has been used to create the new 800 GbE standard. And while the focus is on 106.25G lanes, it's not a hard requirement; the ETC states that this architecture could also allow for larger groupings of slower lanes, such as 16x53.125G, if manufacturers decided to pursue the matter. Focusing on the MAC itself, the ETC claims that 800 Gb Ethernet will inherit all of the previous attributes of the 400 GbE standard, with full-duplex support between two terminals, and with a minimum interpacket gap of 8-bit times. The above diagram depicts each 400 GbE with 16 x 10 b lanes, with each 400 GbE data stream transcoding and scrambling packet data separately, with a bonding control which synchronizes and muxes both PCS's together.
Diving in, the new 800GBASE-R specification defines a new Media Access Control (MAC) and a Physical Coding Sublayer (PCS), which in turn is built on top of two 400 GbE 2xClause PCS's to create a single MAC which operates at a combined 800 Gb/s. Each 400 GbE PCS uses 4 x 106.25 GbE lanes, which when doubled brings the total to eight lanes, which has been used to create the new 800 GbE standard. And while the focus is on 106.25G lanes, it's not a hard requirement; the ETC states that this architecture could also allow for larger groupings of slower lanes, such as 16x53.125G, if manufacturers decided to pursue the matter. Focusing on the MAC itself, the ETC claims that 800 Gb Ethernet will inherit all of the previous attributes of the 400 GbE standard, with full-duplex support between two terminals, and with a minimum interpacket gap of 8-bit times. The above diagram depicts each 400 GbE with 16 x 10 b lanes, with each 400 GbE data stream transcoding and scrambling packet data separately, with a bonding control which synchronizes and muxes both PCS's together.
Non IEEE? (Score:2)
So did a bunch of companies that wanted to do something get a 'no' from the 802 exec committee, leading them to make their own standard to force the IEEE's hand? Or did I misread it?
Re: (Score:2)
The IEEE is slow and (very) expensive to deal with. Both in the Wireless and Wired standards, people have wanted to deviate from them, hence why we have things like InfiniBand and a number of other single-vendor wireless standards for high bandwidth (typically point-to-point or in its own spectrum). IEEE adoption will eventually happen, typically after end consumer devices have already shipped.
Re: (Score:3)
I've spend my years in P802 (I was on the exec committee myself for a bit),802.11, 802.16 and industry groups WFA & WiMax.
It's clear that the industry groups and 802 have well delineated roles. They don't tend to tread on each others domains. Certification, testing, Branding, marketing, lobbying - the industry groups do that. Spec writing, interop, technical consensus building - IEEE 802 does that. The IEEE is slow because what it does is harder. You can build a new thing without consensus, but interop
Re: (Score:1)
Re: (Score:3)
I would have thought that channel bonding should be just below the IP layer and not depend on the underlying transport standard. Then you could bond anything to anything!
As I understand it, what makes bonding tricky is the problem of out of order delivery. While IP theoretically doesn't gaurantee in-order delivery in practice it out of order delivery causes performance issues so network engineers want to avoid it as much as possible.
If you just blindly split packets between two links at one end and recombine them at the other there is a good chance that the packets will not be delivered in the original order due to differences in lengths of the two paths and due to varying
Re: (Score:3)
Not quite.
The IEEE just makes standards. That's it. But to make a standard, people have to come together to debate out the standard and how it would work. And then people need to implement that standard. And then you need to make it interoperate.
Some standards, like HDMI, Bluetooth, and USB are made of industry groups that both
Re: (Score:2)
There are lots of IEEE working groups, but we're talking about P802 here - the owners of the 802.3, 802.5, 802.11, 802.15, 802.16 and 802.1 to bind them together in an interoperable whole. The 'industry' standard here is channel bonding for one of the 802.3 PHYs. Nothing good will come from this being taken out of its natural context.
Rebranded Committee has just rebranded... (Score:2)
... 400G Eth in RAID0 as 800G Eth.
Same signalling, double the lanes.
Voila!
A "new" standard.
Distance is usually a part of a spec (Score:2)
Nowhere does in specify distance. 1 meter? 3 meters?
Re: (Score:2)
Re: (Score:2)
Nowhere does in specify distance. 1 meter? 3 meters?
It's not a specification for the PHY, crudely it's "if you want bolt together 8x100GbE PHYs with some glue to support a 800GbE MAC, this is how you do it".
Re: (Score:2)
It seems this standard is just doubling the lane count compared to 400 gigabit ethernet. Like with 400 gigabit Ethernet one would expect there to be multiple physical layer options with different tradeoffs between range and cost.
Going by the table on wikipedia for 400G it seems we can expect 150m on high grade multimode fiber and tens of kilometers or so on single mode fiber with standard transcievers. I would expect systems with optical amplification will exist to take signals further if needed.
That's nothing! I just unveiled 6.4Tb Ethernet! (Score:2)
Our PCIe NIC is from the ... "switch" ... brand, and requires an arm-thick Ethernet "cable" and has a connector with 8 ... "pins" ... that must be separately connected. :D
Re: (Score:1)
Re: (Score:1)
Re: (Score:1)
Well, that's just spiffy (Score:4, Funny)
Now I have to re-cable my house - again -. I'm not sure this E-mail stuff is worth the fuss.
Re: (Score:1)
Frame size? (Score:3)
What sort of frame size do you use on a nigh-terabit link?
Re: (Score:2)
The same ones you use on gigabit ethernet. The hardware offloads a lot of the work reassembling multiple frames into blocks of data for the host to handle and DMAs it all into RAM.
With hardware offload frame size isn't a big issue any more.
Re: (Score:2)
Meh (Score:2)
They basically glued 2x PHYs together. No coding advancements, no electronic innovations
Re: (Score:2)
It's a bit more complicated than that. See https://hardware.slashdot.org/... [slashdot.org]
Meanwhile in the real world... (Score:2)
...there's still nothing for SoHo applications above 1 GbE. OK, you have the odd ethernet card or even integrated port goind up to 2.5 / 5 / 10 GbE, but no switch, no router.
Re: (Score:2)
...there's still nothing for SoHo applications above 1 GbE. OK, you have the odd ethernet card or even integrated port goind up to 2.5 / 5 / 10 GbE, but no switch, no router.
There are tons of SoHo options for 2.5/5/10GbE. I use a Ubiquiti US-XG-16 in my homelab and its awesome. Cost me $465 on Newegg. You can get them used for less than $325 from time to time on ebay.
Tour of my Home Network:
https://www.youtube.com/watch?... [youtube.com]
Why don't we do these in multiples of 10 anymore? (Score:2)
I kinda miss the days where the new Ethernet standard was usually 10 times faster than the prior standard. You know, like going from 10 megabit, to 100 megabit, to gigabit, and then to 10 gigabit. It made the math for bandwidth upgrades a lot simpler.
Now we have weird standards like 25 gigabit, 40 gigabit, and now 800 gigabit. How about standardizing on 100 gigabit first?
Re: (Score:2)
Re: (Score:2)
Seriously, though. When something like a PCI Express gigabit network card died, it was easy to find a replacement in storage.
When one of those custom 40 Gigabit server modules goes bad... good luck finding a quick replacement unless you're a giant data center and have a bunch of identical servers laying around.
Common standardized parts are a good thing!