Is it just going to require NIC firmware updates? Or will it require upgrading the NICs too?
The move beyond gigabit Ethernet is gathering pace, with a cluster of vendors gathering around the IEEE standards effort to help bring 2.5 Gbps and 5 Gbps speeds to the ubiquitous Cat 5e cable. Members of the MGBase-Talliance include Aruba Networks, Broadcom, Brocade, Delta Electronics, Freescale Semiconductor, Pulse …
Even if it requires replacing the NICs, that's still a lot less troublesome prospect than having to dig around to replace copper. Getting the most speed out of the existing stuff is going to be a pleasing prospect for at least some people.
Maybe a bit of background, El Reg, as to how they're going to achieve this?
Sorry, but by the time they work this out, 10Gb kit will be cheap enough.
At the moment, it's easier - and infinitely cheaper - to LACP a handful of copper or fibre (who the hell runs single fibres or single cables?) and when that's no longer sufficient, a 5Gb Ethernet will only be a small stop-gap at best.
To be honest, they should have come up with something 10 years ago. In the meantime, even ancient Gigabit routers and switches that I have in front of me support LACP. Sure it means an extra cable run, but most places put that in by default, so 2Gbps is a given at the very least. And if you're going any distance, it's 8-core fibre, so that's 4Gbps before you even start.
I bet the cost of configuration plus another cable run, or another couple of GBICs (not that they're called that any more) will be a significantly better investment than any amount of hanging around to replace all your kit with 2.5/5Gbs kit. If you're going to do that, you might as well just wait until 10Gbps is affordable.
10Gb may never run acceptably over some existing cable plant. I've several hundred applications where 2.5/5Gb to the desktop would be a very attractive upgrade, especially if there's some hope of getting the desktop end for "free" with a client refresh.
10Gb has been on the 'will be cheap soon' list for quite some time. It doesn't seem to be getting closer to getting off it.
10Gb is at £100 per card / port. 100BaseT was in the same position, as was Gigabit.
However, who's going to invest similar amounts (which it won't be, as the kit will be harder to push and in lower volumes, and new untested products) in 5Gb when 10Gb is already there?
It's a dead middle ground, and the cost of a few extra Cat6 runs alongside Cat5 runs down 10+ years ago is minimal compared to the upheaval of any method anyway.
The problem is that it will only buy you a couple of years, at 10Gb at best, on old cable, if you're lucky. Chances are that the tolerances will mean that although it's possible with Cat5e, everything has to be new and perfect.
This looks like marketing puff. There is absolutely no content on the MGBase-T Alliance website - could this just be a PR exercise, whilst the real action happens elsewhere?
Is this just a ruse to distract people from where the real action is? There is also the NBase-T Alliance, which purports to do the same thing - deliver 2.5Gbps and 5Gbps over twisted pair. This is backed up by actual work being done at the IEEE - and was reported on by The Reg: http://www.theregister.co.uk/2014/11/11/ieee_turns_crank_on_new_wired_ethernet_standards/
Methinks the MGBase-T Alliance are bandwagon jumpers.
RE: This looks like marketing puff.
I very well may be, but there is s shitload of OLD plant that could benefit from increasing speed.
For the beancounters, replacing NICs and speccing better NICs on equipment refresh may be more cost conscious than ripping up plant, especially IF it involves ripping out plant in leased premises.
No beancounter trying to fund executive bonuses is going to authorize a rip and replace on premises that have a short term left on a lease, with no guarantee that their company will be able to obtain favorable lease terms for a renewal.
Typical short term thinking executive behavior - kick the can down the road.