Netonix Forums

mhoppes wrote:HAHA! What I meant was directed to Chris :) Flow Control shouldn't be needed on the Netonix switches when interfacing between 1Gig and 100meg right? Because of the larger buffer sizes?

Chris advocates the use of flow control on the WISP switch.

Hmmmm OK. I thought I had read somewhere that it wasn't necessary with properly behaving switches and devices?

mhoppes wrote:HAHA! What I meant was directed to Chris :) Flow Control shouldn't be needed on the Netonix switches when interfacing between 1Gig and 100meg right? Because of the larger buffer sizes?

No you still need it but the larger buffers help alot as we do not have the lop sided speeds.

It all depends on how much data is being sent to the 100M device and how quickly the device can push it out. The worse the wireless connections the more you need FC.


I have made SEVERAL posts telling people to turn FC on when using 1G and 100M devices on the same switch.

Larger buffers can only do so much, we are dealing with a half duplex wireless link with an unknown ability to to pass data based on noise or poor links having tons of wireless retries which causes your buffer to over flow.

Now normal TCP protocol should handle this but for some reason the airMAX radios "seem" to handle something differently, not sure? We need to investigate what the airMAX radios are doing when too much data hits them bound to a CPE radio. Since the CPE radio is not the final destination TCP protocol may not be able to deal with it properly thus the need for Flow Control.

When we get time we want to look and see what the airMAX radios are doing when their buffers over run. Not sure if UBNT could have interjected something there and did not? However if we find something we will share our findings to Ubiquiti because we do like them even if they do not like us.

But for now if you have 1G and 100M devices turn Flow Control on for the 1G up link port and the 100M radio ports.

sirhc wrote:When we get time we want to look and see what the airMAX radios are doing when their buffers over run. Not sure if UBNT could have interjected something there and did not? However if we find something we will share our findings to Ubiquiti because we do like them even if they do not like us.

I have a strong suspicion they already know, but just aren't saying -- and I'm not even the paranoid type.

Scott

sirhc wrote:Thats because you use automobile inverters Matt.

I have told you to buy Tripp-Lite Inverters many times, use a real inverter!

http://www.ebay.com/itm/TRIPP-LITE-APS7 ... 2a41e2ef21

I power my entire tower off these bad boys!
Cisco 2951
WS-24-400A FULLY loaded with radios

Exactly what I use at every tower! Sealed AGM 75 or 100Ah batteries, up to 4 in parallel, and I get 1-2 days runtime (depending on temperature too). Also, I have at least 1 or 2 per year, I have to run out to because the WHOLE TOWER SHUTDOWN due to "overload" on the inverter. A simple switch to "off" then back to "auto" resets it. Not sure what causes it, maybe a power surge, though I've installed all the surge protection one can think of and I still have a tower shut down (and anything else backhauled behind it) because of this stupid glitch.

HOWEVER, strip out this bulky inverter (that's a PAIN to cram into a outdoor NEMA enclosure) and replace with a DC-powered POE switch that can power AirFiber, and not only will I have a less-messy box, but I bet the runtime will increase 50% or more. Not to mention I've never had a lock-up issue at my sites running 100% pure DC, so I'd say reliability increases as well, not having your backup inverter trip into an overload protection state, from a surge.

Last but not least, is this "WISP Switch which is "SMART" DC" available yet???? I need to order one for every tower we have, asap!


-WISPWEST

We power every tower from four 100ah AGM batteries wired to 24v charged by a Samlex charger. Currently use TS POE's and MT 750UP's (for port isolation) along with MT routers all powered via DC. Upconvert to 48v for AF when needed. A tower with no AF on it will run for 5-6 days without grid power. Don't need to bother with inverters - converting from grid to batter back to inverter just doesn't make sense.

We are really looking forward to the DC Wisp switch to replace all of our TS and MT switches. With a few more refinements on port isolation on the WS and the DC version, we'll have a rockin solution.

Have checkbook out....just need switches now. :-)

Hi Chris - Are you planning a 8 port non-RM and a 12 port RM DC switch or just the 12RM? I am planning 1/2 dozen new tower boxes and this info would be helpful.
Rob

rkelly1 wrote:Hi Chris - Are you planning a 8 port non-RM and a 12 port RM DC switch or just the 12RM? I am planning 1/2 dozen new tower boxes and this info would be helpful.
Rob

There are several posts explaining what we are doing, some even have pictures. I know we only have 100+ users on this forum but you all seem to ask the same question and there is only me to answer the same question 100+ times.

We are soon releasing an 8 port AC switch that is almost the same size as the TOughSwitch but it does eveything our larger switches do, same software, same everything, powers airFIBERS....Same features SMALL chassis.

Then later this spring we are releasing a 12 port plus 2 SFP ports, same capabilities as a WS-12-250A but in the se chassis as our 8 port switch.

We will also release a SMART DC version of the 8 port, 12 port mini, and WS-12-250A that can accept from 9 to 60V DC.

Here is a link to one post describing these viewtopic.php?f=8&t=275#p1119

Thanks. I knew all of the stuff except on the DC side. Must have missed it...