Please make a Gigabit Reverse PoE switch, with 14-24 PoE in ports, 1-2 PoE out ports, and a SFP(+) port

Hardware MikroTik hardware general
1

Please, please make a Gigabit Reverse PoE netpower CRS switch, with 14-24 PoE in ports, 1-2 PoE out ports, and a SFP (SFP+ would be really awesome) port.

Why?

Because this would be THE product for MTU scenarios with no shared telecom room where we are relegated to being a weatherproof box on the back of a building, with limited to no power options.

Here’s how it would work:

Mount a weatherproof box (which could be the switch itself) on the outdoor wall. Run fiber to it (using a SFP GPON optic like the UF-Instant or the like, a standard Active Ethernet SFP(+), or PoE out to power a UF-Loco or UF-Nano). Each tenant gets outdoor rated CAT5e/6 from their router to the switch, with a PoE injector.

The benefits:

With multi reverse PoE, I don’t have to worry about a specific customer being the power source and bringing the whole complex down when they trip over their PoE brick (or just move, or don’t pay their electric bill, etc).

With Gigabit, I can sell our fiber plans (which start at 200M and go to 1G, which is why the Netpower 15P, which is a really cool idea, just doesn’t cut it with 100M ports).

With a SFP+ port, we have options for when people start buying a gigabit plan on these. I’m hoping we (the WISP turned FISP community) could lean on Ubiquiti to create a UF-Instant+, a SFP+ version of their SFP ONT, capable of up to 2.5 Gbit down and 1G up to put in the SFP+ port. OR, Mikrotik could make such a thing, and I’d happily buy it, if you worked with Ubiquiti to make it compatible with their PON platform. Even without that, SFP+ gives the option to go to 10G active ethernet if I get a couple of gigabit plans sold.

I think you would be surprised how much people would pay for an option like this. I think you would be able to get to $500 (for a CRS line switch) before we would really even notice. Maybe even higher. (I’m not the money guy, but hundreds of dollars for GPON CPE that serves 1-5 customers is not rare unless you are in the world of Ubiquiti PON.) Not having to find a power source, hire an electrician (and thus get lots of consent from a complex manager) saves us so much time, money, complication, and opens up so many options on many apartment complexes.

2

So basically a netPower 15FR with Gigabit and SFP+'s?

I can agree that releasing a 10/100 switch in 2019 is, a bit conservative. However I personally chosen to only have 100/100 at home. Haven’t seen any need for anything above that.

3

Yes, exactly. Without dropping the reverse PoE capability.

Right, but when you are trying to compete with Charter/Spectrum (large cable provider in the US covering several states) that is running DOCSIS 3.1 and selling 100, 400, and 900 Mbps plans, you have some good reason to start at 200M and make it just a little cheaper than their long term 100M rate. Basically, you end up getting thin margins on payback of the nearly fixed initial cost of deploying fiber, but can’t raise the price because of competition. But normally serving 100M vs 200M on GPON is a minuscule (bandwidth) cost increase, with no increase in labor or materials. (Until we start talking about switches with 10/100 ports!) So you don’t try to compete on lowest price, but instead best value: better upload & download speeds for about the same price. Even if the customers, 99% of the time, don’t do all that much with the more available bandwidth.

4

Maybe Mikrotik will continue with an Gigabit-version.

In fact, I see that problem in Sweden sometimes. Cable providers oversell capacity for dirt cheap on their lousy old long-payed-off coax-networks. Their network is often plauged with groundloops, weather-influence, irregular ping-times and regular disturbances and downtime, but they are indeed sometimes the “cheapest per megabit”, in downstream at least. Some even sells 1200/30 or something.. it’s ridiculous and useless…

Then it’s of course hard to be an honest FTTH/FTTB provider and charge a realistic price, even if the connection is ten times more reliable and symmetrical. I hope that people will come to their senses in time.

5

I found a SFP+ ONT on a stick optic:

https://www.aliexpress.com/item/4001075949216.html

Haven’t actually bought one and tried it. But it is an example of something you could use with an SFP+ uplink port.

6

But, it needs be compatible with MikroTik devices and with different OLT’s.

Regards.

7

Many existing devices with multiple PoE outputs (hex poe, omnitik, powerbox) could be redesigned fairly easily to support reverse PoE-in on the same ports normally used for PoE-out.
Currently, if you apply power to a PoE output, nothing happens because of a reverse-biased series blocking diode.
By adding a diode from each PoE-out to power supply input, you convert PoE-out to reverse PoE-in (as long as PoE-out is disabled on that port).
One possible issue is power budget, with gigabit ports the ethernet transformer must be rated to pass enough DC current, so all ports must have the same sized transformers as the one PoE-in port (usually ether1). Sometimes this means you have to use 48V power supplies, and powered devices that accept 48V.
Note that 10/100 ports don’t have the issue (current still limited as we don’t want RJ45 connectors to melt, but no transformers in spare pairs), so perhaps a hybrid device with some 10/100 and some gigabit ports would be an acceptable compromise. CRS318-1Fi-15Fr really should at least have one gigabit PoE uplink port, currently I work around this by using a copper SFP and separate gigabit PoE injector (which takes power from the PoE-out port, and ethernet from the copper SFP) to connect the LHG60.