Re: Typical power usage of GS752TP

MikeDC · ‎2020-11-26

Can anyone tell me what the minimum and typical power consumption would be for this switch? Each watt consumed equates to about $2 per year, so if it uses 100W with some mild traffic then that's $200 per year. If it uses 30W then it's a much more reasonable $60. Cisco include this sort of information for their stuff but all I can find for Netgear is maximum power usage. For me the POE side of things I consider as zero power as the devices are currently powered anyway. I'm thinking the PoE could actually save some power as I would have a single power supply instead of a bunch of smaller ones.

There was a previous thread on this here but it was marked solved without any figures being given.

https://community.netgear.com/t5/Smart-Plus-and-Smart-Pro-Managed/GS748TP-Switch-How-much-Power-does...

schumaku · ‎2020-11-26

Data Sheet GS516TP, GS728TP, GS728TPP and GS752TP

POWER CONSUMPTION (WHEN ALL PORTS USED, LINE-RATE TRAFFIC AND MAX POE)
Worst case, all ports used, line-rate traffic, max PoE 512.8W

PoE budget (PSU/Passthrough) 384W

So with all links at line rate and no PoE but including the PoE power supply overhead it's max. 128.8W

Now asuming the power supply is an 85% efficiency unit, PoE availability alone does require 452W effectively (68W loss by overhead!), the 100% load of the switch does make about 61W for the switch alone.

So we Netgear is about in the same range as a Cisco Catalyst 2960X-48 at 100% load (Cisco Catalyst 2960-X and 2960-XR Series Switches Data Sheet, p.29). 100% to 10% load is about 2% less power only, idle 0% load about 24% so I estimate some 46W for your switch model.

When we compare the 600 USD for a GS752TP to the 3000 USD for the Cisco ... this is a good part for you pay for the additional information they provide. So you can burn much more power on the Cisco until the cost are exceeding the hardware 8-)

Oh CPC does show 60.9W FWIW:

View solution in original post

schumaku · ‎2020-11-26

Impossible to answer, well at least in the non-acaemical way. Buy a unit, and set-up some test scenarios.

The core is up 24*7, some power supply overhead is there 24*7, every 10M100M link does require some power, every GbE link established (idle) does require power, Link power requirement an vary because of EEE/Green Ethernet when using shorter connections, ... and then the variance on how the switch does has to handle throughput where the core and PHY will require more power, lack of mechanical work creating more heat, requiring the fans to spin on a higher speed requiring even more power ... Sponsor a student doing some academic work on it to develop the "magic" formula.

But then, the formula requires exact input of your use case. So what does the community know how your switch, your network is used?

MikeDC · ‎2020-11-26

It's not that hard to answer, Cisco give a few example use cases and give a warning that it can vary. It's not likely to vary that much. A figure with nothing at all connected would be easy to measure and give a good starting point. Then a figure with say 8 ports connected with some low traffic would be good. I think if someone measures 50W with 8 ports connected then it's a very good indication, it's not like it's going to jump to 150W for another user with 8 ports. On the other hand if it uses 150W at idle that is a good indication it's power usage will cost more than the switch over its lifetime.

schumaku · ‎2020-11-26