Chip supply problems might mean Wi-Fi 6E is skipped over for Wi-Fi 7, says analyst

Supply chain woes with Wi-Fi 6E products could see organisations miss on deploying network kit with the new standard and instead wait on availability of Wi-Fi 7 equipment expected next year, says Dell'Oro Group. Wi-Fi 6E builds on Wi-Fi 6, itself only a newish standard, by adding support for frequencies in the 6GHz portion of …

[Anonymous Coward]

> although manufacturers might have launched Wi-Fi 6E devices, such products are often either not available, or they are in very limited supply.

It didn't help that most Wifi 6 gear inexplicably launched with GbE on the backend, despite sporting multi-gigabit-capable radios. Even now the newer 6E devices are finally shipping with the appropriate-sized 2.5GbE, the prospect of procuring enough 2.5GbE switches to actually take advantage of that is enough to put everyone off for the next year or two. There just isn't the kit.

[Anonymous Coward]

I haven't understood why 2.5 is even a thing, should go straight to 10 as 2.5 isn't a great increase to reflect the current stepping of speed. I can see 2.5 in places like RPI's or other small SBC's, but on anything else it's simply too little of an increase to care about when there's now 10GB/s storage devices along with CPU's, buses and memory that can handle much more. 2.5gbe kind of looks like 56/k dial-up, the black sheep of all your interfaces.

Another "standard" nearly skipped over is PCIE 4.0. 5.0 is now being packed and 6.0 is now in sight of large'ish data players as it's standardized.

"... there is no guarantee that upcoming Wi-Fi 7 products will not be affected by the same issues of semiconductor supply shortages..."

Sure, but that simply means everything is equally hard to acquire so why not build 7 instead of 6E?

[Anonymous Coward]

2.5G and 5G work on cat 5e cables - as long as they are good enough - and were introduced to ease the requirements of 10G networks (which needs cat 6/6A cables, or fibre) - they may also deliver cheaper low-end devices, as 10G ones didn't became less expensive enough yet, and also 10G network cards and devices run hotter.

There are also many desktop motherboards coming with 2.5G NICs, but not many with 10G ones.

That could change, of course, if 10G hardware becomes cheap enough soon to become mainstream and make 2.5/5 Ethernet far less interesting.

Anyway as far as I can see WiFi 6 is becoming mainstream only now - especially since ISPs started to deliver WiFi6 CPE in the last few months, before it was still mostly WiFi 5. And user have also to replace a lot of their devices to get WiFi6 ones. WiFi7 in 2023? Looks quite early to me.

[Giles C]

And by the time WiFi7 devices start going on sale someone will want to buy WiFi8.

I’m all for advancement but sometimes the hardware manufacturers of the APs and the interfaces onto devices maybe should coordinate things a bit better.

Most companies don’t replace everything on an annual cycle - why would they need to….

[Yet Another Anonymous coward]

How will you feel sending email in 2022 using 2021 era WiFi ?

[Paul Crawford]

More affluent?

[Marcelo Rodrigues]

Quite well, thank You

"How will you feel sending email in 2022 using 2021 era WiFi ?"

I'm sending email using 2015 era WiFi... An I will keep doing this, as long as it delivers enough bandwidth - I use WiFi only to mobiles and Kindles, everything else is cabled.

[Sandtitz]

I'd also like to add a critical reason for 1/2.5/5Gbit connections: they can provide PoE power. Wikipedia tells me that 10G PoE standard exists but I can't seem to find any products.

LACP is your friend as someone put it. The other port doesn't have to be PoE port.

Another reason the 6E hasn't been deployed is you need to replace the clients as well.

[batfink]

Yes desktop mobos will be coming along with 2.5G NICs, but as usual I have to ask "why?". What is it that people are doing on their desktop that requires a 2.5Gb LAN link?

I can understand this in a specialised work environment, but if you're doing the kind of number-crunching that requires that kind of bandwidth, you're not going to be doing it on your average desktop.

As for home use: wtf? So your backup finishes in half an hour instead of an hour - do you care? And even if you're trying to download all the world's Pr0n you're still not going to be able to watch it at 2.5Gb/s...

[Anonymous Coward]

coming along with 2.5G NICs, but as usual I have to ask "why?"

Maybe no good reasons but as long as they become available some people will want 2.5G networks to use them.

Here some ISPs are now offering 2.5 and 5 Gb/s FTTH connections - and people (without real good reason) wish to "exploit" those speeds. That's after all what device makers look at - "there's demand, let's fulfill it - if we can stick a premium price on it the better."

Even WiFi 6 speeds that with high-end devices can get past the 1Gb/s speed may be not really needed - but they are here.

I'm not going to scrap my 1Gb/s LAN even if I can get a 2.5+ FTTH connection, but others look willingly to do.

[John Brown (no body)]

Re: coming along with 2.5G NICs, but as usual I have to ask "why?"

"I'm not going to scrap my 1Gb/s LAN even if I can get a 2.5+ FTTH connection, but others look willingly to do."

I remember doing the same with 10 10Mb/s co-ax network. Switches still being expensive, I got a hub with a BNC connector so those devices capable would get RJ-45 cables and the rest, such as the laser printer, would stay on the co-ax network. It took a few years before that last bit of co-ax got retired in favour of 10Mb/s Ethernet. More years again to migrate to 100Mb/s Ethernet. I have a Gb switch now, but I think most devices on my network are "only" 100Mb/s capable, just the cable modem, it's cable to the firewall, and the upgraded NICs in the firewall.. One day, I'll start replacing cables to the rest of the network, but I'm not seeing a compelling reason yet.

[DoContra]

Future proofing mostly (1Gbps NICs weren't unheard of in PCs in the early '00s, and by the late '00s/introduction of the Athlon64/Core2 parts they were becoming the standard in all but the most cost-optimized desktops), although with SSDs easily hitting 1Gbps+/100MiB/s+ of throughput (with high-end parts hitting ~500MiB/s+) it makes sense to beef up your network link. And the entire point of the 2.5/5 Gbps standard is low cost (2.5Gbps is certified for 100m* on Cat 5e cabling, and 100m* on Cat 6 cabling for 5Gbps, vs 10Gbps that requires Cat6a for the 100m*) and compatibility (these standards should interoperate with 10/100, while I *believe* 10Gb doesn't). My only disappointment is that integrated NICs aren't 5Gbps capable (I thought the only difference between 2.5Gbps and 5Gbps on the equipment end was the Ethernet jack, but I guess I'm wrong).

I've had to buy a couple L2 switches for my workplace, and the only reason I went for 1Gbps was that I couldn't find any 2.5/5Gbps products in my local version of eBay (and the ones I got -- Mikrotik CSS -- are a steal if you can live with web-only admin and no 802.1X)

*: Testing protocol AFAIK is Equipment <-> 5m stranded patch cord <-> 90m solid cable <-> 5m stranded patch cord <-> Equipment (straight cable runs or solid cable patchcords -- SOP in South America -- should allow a bit more distance)

[-v(o.o)v-]

Cat5e supports up to 10 meters 10GE just fine, using them on a half rack worth of servers (the rest are single-mode).

[Chris 211]

2.5G, because you have to support the existing CAT5 structured cabling which NOBODY is going to replace just for WIFI. A few installs I have done the existing structured cabling for 1000s of APs barely supports 2.5G. Yes you can bond the ports if the AP has two ports but then you need to double your switches.

[Throatwarbler Mangrove]

LACP is your friend.

[Anonymous Coward]

LACP is the underlying protocol of LAG

As long as you have three ports to connect each device pairs - which is not always the case. And a 24-port switch becomes an 8 port ones this way....

[mattaw2001]

I'm not sure LAG/LACP is the answer here as the load balancing to and from a single client cannot use more than one physical link, which would max at 1gps.

Although there are some new(er) balancing algorithms that work on a port & ip basis for different UDP and TCP ports if there are two in use.

[Kevin McMurtrie]

WiFi infinity

Customer: What I really want is low packet loss and jitter through walls so I can video conference

Wi-Fi Alliance: Line-of-Sight speeds are going to be faster than your cable!

[Korev]

Re: WiFi infinity

Very true Kevin, have a pint

[Disgusted Of Tunbridge Wells]

Re: WiFi infinity

I get terrible signal in my garden - enough for my phone to cling on to, not enough to actually work.

The benefit of eleventy-ghz connections is the signal won't make it at all so my phone will just use the mobile data.

I'm not sure that's the improvement they were aiming for, but it will fix my use case.

[Charlie Clark]

Re: WiFi infinity

It's one of the "benefits" of LTE / WiFi convergences. Phones will know already use WiFi connections for calls if they're available with the reason being that WiFi is likely to better indoors than the local LTE signal. The next step is for even more integration so that the networks can use more of our gear for their signals. So get on and install that WiFi-repeater in your garden!

[Disgusted Of Tunbridge Wells]

Re: WiFi infinity

> So get on and install that WiFi-repeater in your garden!

What's the point when Android doesn't hop around wifi connections aggressively enough?

[Charlie Clark]

Re: WiFi infinity

It doesn't seem to do so here: two wifi repeaters in the house. Never have problems with my Android devices but all the Apple gear has to be cabled for when DFS kicks in and 5 GHz channels get disabled at the behest of the local airport.

[Roland6]

Re: WiFi infinity

>What's the point when Android doesn't hop around wifi connections aggressively enough?

Many PC WiFi cards also don't hop around, it would seem they have been configured on the assumption that the typical home only has one WiFi AP, so the need is to hang on to that connection. It will be interesting to see whether mesh friendly adaptors are any better.

I've found, in the absence of settings on the client, adjusting the tx power level on your AP's and lowering their offloading threshold can improve matters. One PC (with a TP-Link WiFi adaptor) I had to turn off an AP for 10 minutes before the PC would roam to the closer AP with a stronger signal and remain associated with that AP when the old AP was restarted...

[Chris 211]

Re: WiFi infinity

WIFI repeater! Wash your mouth out! We dont do repeaters.

[Roland6]

Dell'Oro Group are obviously trying to talk up the next "new shiny"

Given the current situation, I actually see little real demand or need for anything beyond WiFi 5 for the next few years, since all those new laptops purchased during lockdown will only be WiFi 5 / 802.11ac 2x2 at best...

[Charlie Clark]

Re: Dell'Oro Group are obviously trying to talk up the next "new shiny"

True, but the network infrastructure is likely to have a longer refresh cycle. So, if as we are, you're looking to upgrade it this year, 6 is probably the way to go. 6E sounds like it was rushed out of the door to trumpet higher theoretical transfer rates through more unlicensed spectrum, when many of the advantages of ≥ 5 are down to better management on the way to a cellular network: more cells using less power means better coverage with less contention.

[Roland6]

Re: Dell'Oro Group are obviously trying to talk up the next "new shiny"

>more cells using less power means better coverage with less contention.

Not convinced, but then over 10 years back did an auditorium with Aruba kit and subsequently used Xirrus AP's for higher density office deployments.

For one client the limiting factor now isn't the WiFi, but the fact their office doesn't have acoustic cubicles/spaces and hence sets a practical limit on the number of concurrent Zoom calls from iPads....