Where are the cheap 2.5G switches?

[Modred189]

Its been a decade of cheap gigabit consumer switches. You can get them anywhere, and every chinesium brand will sell one for $15.

 

But where are the cheap consumer 2.5GB/s switches? With the proliferation of consumer/prosumer NAS devices and IoT DIY spaces, why haven't we seen a proliferation and cheapening of good networking gear for less? I can't find a reasonable 2.5GB/s switch for unde $115 ish?

[Biohazard777]

41 minutes ago, Modred189 said:

With the proliferation of consumer/prosumer NAS devices and IoT DIY spaces, why haven't we seen a proliferation and cheapening of good networking gear for less?

IoT devices don't need much bandwidth, 100Mb/s is usually more than enough for those... so not sure why would an increase of IoT devices drive the cost down for 2.5 Gb/s or 10 Gb/s switches.
As for NAS devices, sure I can see the correlation there... but I'd be surprised if just 1% of all PC/laptop users have a NAS.

As to when we will see cheap 2.5 Gb/s switches, about the same time when even entry level motherboards and laptops start shipping with 2.5 Gb/s NICs.

[Alex Atkin UK]

4 hours ago, Biohazard777 said:

As to when we will see cheap 2.5 Gb/s switches, about the same time when even entry level motherboards and laptops start shipping with 2.5 Gb/s NICs.

Precisely, and lets not forget its taken over a decade for Gigabit switches to get as cheap as they are.  Its practically prehistoric tech at this point.

[SVThuh]

The deals are out there.  I just picked this one up and it was indeed open box, but it was brand new and unused..

 

[NobleGamer]

On 9/20/2022 at 8:25 PM, Biohazard777 said:

IoT devices don't need much bandwidth, 100Mb/s is usually more than enough for those... so not sure why would an increase of IoT devices drive the cost down for 2.5 Gb/s or 10 Gb/s switches.

 

I don't think OP was suggesting that IoT's bandwidth should cause 2.5gb switches to be lower price or higher adoption.

 

Rather, the point is more like "Hey there's a bunch of other tech that uses networking that has gotten much cheaper over time - So why hasn't a similar trend happened with networking with ports over 2.5gbps?"

 

Absent someone from a big consumer networking company actually providing an answer from a holistic industry insider perspective and not an armchair analyst outsider, I suppose what we're left with is that the use cases for 2.5gbps aren't large among consumers and cost of whatever is needed to support 2.5 (the ports, switching throughput, etc).

 

Its tough to look at from an enthusiast/power user perspective because we can think of a bunch of use cases, but we have no idea how many people actually apply to the use cases: NAS, supporting backhaul to wifi 5 or higher, not bottlenecking 1gbps or higher internet, etc.
 

On 9/20/2022 at 8:25 PM, Biohazard777 said:

As to when we will see cheap 2.5 Gb/s switches, about the same time when even entry level motherboards and laptops start shipping with 2.5 Gb/s NICs.

Its interesting to contrast that with the theory that there's not enough 2.5gb usage/demand, because if that theory were true, then 2.5gbps ports in laptops & mobos would have to lead before cheaper 2.5gb switches - Not necessarily at the same time.

[Modred189]

On 9/22/2022 at 11:31 PM, NobleGamer said:

 

I don't think OP was suggesting that IoT's bandwidth should cause 2.5gb switches to be lower price or higher adoption.

 

Rather, the point is more like "Hey there's a bunch of other tech that uses networking that has gotten much cheaper over time - So why hasn't a similar trend happened with networking with ports over 2.5gbps?"

 

Absent someone from a big consumer networking company actually providing an answer from a holistic industry insider perspective and not an armchair analyst outsider, I suppose what we're left with is that the use cases for 2.5gbps aren't large among consumers and cost of whatever is needed to support 2.5 (the ports, switching throughput, etc).

 

Its tough to look at from an enthusiast/power user perspective because we can think of a bunch of use cases, but we have no idea how many people actually apply to the use cases: NAS, supporting backhaul to wifi 5 or higher, not bottlenecking 1gbps or higher internet, etc.
 

Its interesting to contrast that with the theory that there's not enough 2.5gb usage/demand, because if that theory were true, then 2.5gbps ports in laptops & mobos would have to lead before cheaper 2.5gb switches - Not necessarily at the same time.

Yea, the chicken and the egg scenario is real here. That and the perceived complexity of NAS systems resulting in reliance on cloud storage. 

 

It just seems like 2.5G tech would be a big selling point, given the wifi speed races, and there being so little price competition is disappointing.

[jimm_eh]

Years ago, hard drives suddenly went from 2MB to 8MB onboard cache after a very long time of being stuck at 2MB.  Of course the HDD companies pitched it as an upgrade, but my suspicion was that this happened not from HDD customer demand, but because the other 99% of the memory market had moved on to larger sizes and the memory companies just stopped making anything that small.

Gigabit is similar, except that the customer base that **isn't** demanding the extra speed has been the 99% of the market. So it isn't likely that supply-side factors are going to force the upgrade here. Perhaps if greater-than-gigabit broadband becomes common enough to induce sufficient numbers of end users to start picking up 2.5gbe parts, then the shift will happen.  I strongly suspect that at least the 2.5Gbe parts today are already close to gigabit parts for cost and power consumption.

 

[Alex Atkin UK]

11 minutes ago, jimm_eh said:

I strongly suspect that at least the 2.5Gbe parts today are already close to gigabit parts for cost and power consumption.

Well, the Intel i225 NIC chip is only around $2.71 in bulk.  This explains why its being used all the over the place, its a cheap bullet point to add to your product.

 

Unfortunately demand for switch chips is going to be a lot lower and thus the bit to connect it all together is still kinda pricey.

 

I wouldn't be surprised to see Gigabit ports on routers still when WiFi 7 comes out, capable of doing 2.5Gbit over WiFi.  There's still going to be such a massive stock of Gigabit switch chips lying around, its an easy cost saving and they can charge stupid money for the high-end routers that hopefully WILL be 2.5Gbit across all ports at that point.

 

Still, its those 5 port 2.5Gbit switches that ultimately will end up embedded in routers.  Four ports for LAN, 1 for SoC then the second SoC port for WAN.  Its just a matter of time.

[Donut417]

5 hours ago, Modred189 said:

t just seems like 2.5G tech would be a big selling point, given the wifi speed races, and there being so little price competition is disappointing.

The biggest selling point for 2.5 Gbps is faster internet for the average person. Some cable companies can do 1.2 Gbps down currently on their network. AT&T offers 2 and 5 Gbps Fiber, but only in very limited areas. You have a better chance of having a cable provider in the US than Fiber. So really there is not much of a market outside of that. 

 

As far as the WIFi speed races go. WiFI speeds are not guaranteed. You will never get the speeds advertised, not in the real world. Even WiFi 6 will struggle to give you Gigabit speeds. Granted we are getting closer to that reality but the Airwaves can vary area to area. Where Ethernet is guaranteed as far as speeds go. 

[NobleGamer]

19 hours ago, jimm_eh said:

Perhaps if greater-than-gigabit broadband becomes common enough to induce sufficient numbers of end users to start picking up 2.5gbe parts, then the shift will happen.  I strongly suspect that at least the 2.5Gbe parts today are already close to gigabit parts for cost and power consumption.

 

15 hours ago, Donut417 said:

The biggest selling point for 2.5 Gbps is faster internet for the average person. Some cable companies can do 1.2 Gbps down currently on their network. AT&T offers 2 and 5 Gbps Fiber, but only in very limited areas.

 

Agreed that it's gonna take much more 1+ gbps internet usage, but unfortunately I can think of yet another reason holding back 2.5gbps ports: Why have 2.5 ports when link aggregation is "cheap"?

 

My ISPs cable modem/router (<1gbps, maybe DOCSIS 2.x) and even my cheap TP Link 8 port managed switch both do link aggregation. Its not useful for me with 400mbps or even 1gbps internet, but I can imagine that many consumer wifi APs and Routers do or could easily tack on Link Aggregation if they have 2+ ports.

 

I also agree that WiFi 7 and faster may help 2.5gbps port demand some, but that case is kinda kneecapped by the fact that the 1gbps+ theoretical throughput of these standards is often at shorter distances & less walls via 5 GHz or higher, and thus not intended for 1gbps+ at a distance of more than one room or signal absorbing/blocking materials (like concrete).

[Donut417]

1 hour ago, NobleGamer said:

My ISPs cable modem/router (<1gbps, maybe DOCSIS 2.x) and even my cheap TP Link 8 port managed switch both do link aggregation

Most likely Docsis 3.0, I don’t think any US providers allow Docsis 2.0 stuff on their network. I know that several Docsis 3.1 modems had multiple Ethernet ports for link aggregation, but it explicitly had to be supported by the cable provider. 

[Alex Atkin UK]

13 hours ago, NobleGamer said:

Agreed that it's gonna take much more 1+ gbps internet usage, but unfortunately I can think of yet another reason holding back 2.5gbps ports: Why have 2.5 ports when link aggregation is "cheap"?

Link aggregation isn't cheap for the end user who is using the routers switch ports or unmanaged switches.

 

Getting two 1Gbit links into a 1Gbit switch only allows your WAN speed to be spread over multiple clients, each client is still locked at Gigabit (unless you use a 2.5Gbit managed switch).  Even if you then aggregate at a client or two, you are bottle-necked on single-streams and random load balancing.

 

I actually did test two Gigabit ports between pfSense and my switch, how often traffic would end up entirely down one link was annoying.