airMax - aggregation - priority

[Apogee]

Anybody know how the airMax priority affects aggregation settings?

It seems to me that they should be adjusted based on priority since the time slice to xmit frames is different based on priority and active stations. Does airMax adjust agg settings automagically when needed? Does fragmentation occur if the aggregate frame is larger than the available airMax slot to xmit it?

Anybody have the scoop on this?

[esseph]

Anybody know how the airMax priority affects aggregation settings?

It seems to me that they should be adjusted based on priority since the time slice to xmit frames is different based on priority and active stations. Does airMax adjust agg settings automagically when needed? Does fragmentation occur if the aggregate frame is larger than the available airMax slot to xmit it?

Anybody have the scoop on this?

Good question

[Apogee]

Nobody knows? Not even the ubi engineers?

I did a few tests last night and it looks like aggregation may be linked to the metrics|performence issues I and others have been reporting...so far. I won't be able to tell for sure until until ubi explains the relationship with am.

[Apogee]

I've been fooling with aggregation settings all morning and was able to increase bandwidth across my entire network...up to double the previous bandwidth per device by NOT USING THE DEFAULT 32/50000!

The answer to the op would be important information for Ubiquiti to share with us!!!

[esseph]

I've been fooling with aggregation settings all morning and was able to increase bandwidth across my entire network...up to double the previous bandwidth per device by NOT USING THE DEFAULT 32/50000!
The answer to the op would be important information for Ubiquiti to share with us!!!

What settings are you using? It would make more sense that the size be a multiple of the wireless mtu and not some flat number.

[rhauf]

Oh things are getting interesting? its like an investigation of airmax magic sauce and ways to make it tastier.

[esseph]

Something else to think about is the amount of buffer space spread across the time slots, and the number of CPEs.
Aggregation would be less efficient as you spread timeslots out among more CPEs.
Also, there is a possible problem as you queue up frames to spread out among those CPEs that you run out of buffer space.
In a pure PtP scenario though, you would (in theory) max out the number of frames (32) and the size of those frames as a multiple of the wireless MTU (2024). (64768 byte packets).
I just tested this and got a small performance bump over the flat 50000 byte frames, and slightly lower latency. (This is a non-airmax scenario)
Wonder why that isn't the default?

[Deleted Account]

I've been fooling with aggregation settings all morning and was able to increase bandwidth across my entire network...up to double the previous bandwidth per device by NOT USING THE DEFAULT 32/50000!

What changes did you make? If you have poor links, using aggregation may perform worse (if a smaller packet is dropped it's okay, but if larger packets are dropped, more data is lost -- nothing magic here).

What did you find? Lowering aggregation gave you better performance?

-Matt

[UBNT-Edmundas]

Anybody know how the airMax priority affects aggregation settings?

It seems to me that they should be adjusted based on priority since the time slice to xmit frames is different based on priority and active stations. Does airMax adjust agg settings automagically when needed? Does fragmentation occur if the aggregate frame is larger than the available airMax slot to xmit it?

Anybody have the scoop on this?

Aggregate size is adjusted depending on link conditions and what is specified in UI is the maximum. Lower priority CPEs get less slots but aggregation is not dependent on priority but on link conditions.

-Edmundas

[aragon]

Aggregate size is adjusted depending on link conditions and what is specified in UI is the maximum. Lower priority CPEs get less slots but aggregation is not dependent on priority but on link conditions.

Is aggregation reduced/disabled for packets marked for realtime/voice priority with DSCP?

[UBNT-Edmundas]

Is aggregation reduced/disabled for packets marked for realtime/voice priority with DSCP?

Yes. These packets will be less aggregated.

-Edmundas

[esseph]

aragon: Are you aware of the packet markings that AirMax pays attention to btw?

wiki.ubnt.com/Main_Page/QoS_DSCP/TOS_Mappings

[rhauf]

oh man, now i want to assign lowest QoS priority to video (netflix) traffic.. how can i do that?

[esseph]

oh man, now i want to assign lowest QoS priority to video (netflix) traffic.. how can i do that?

LOL really? That's not a good solution. You want to QoS it higher than a lot of traffic, but LIMIT the maximum bitrate, so the customer is forced to change their streaming bitrate.

That way they have buffer-free streaming that doesn't harm your infrastructure(unless all of your users are watching netflix!)

[rhauf]

currently, nobody is having any problems streaming netflix or other video sources.
yes your solution is even better. now how to go about doing that hmm.

[Bean_Planter]

Netflix user education:
technologizer.com/2011/06/23/got-bandwidth-caps-netflix-has-you-covered/
Do it on the router (like the Mikrotik):
wiki.mikrotik.com/wiki/Limiting_a_user_to_a_given_amount_of_traffic_II
Subscribe to a very inexpensive service that automatically does the router thing for you:
www.dynamicbandwidthshaper.com/
'hope this is helpful in some way.
Cheers,
Gene

[esseph]

Gene you are such a stud

[Bean_Planter]

[Apogee]

What changes did you make? If you have poor links, using aggregation may perform worse (if a smaller packet is dropped it's okay, but if larger packets are dropped, more data is lost -- nothing magic here).

What did you find? Lowering aggregation gave you better performance?

-Matt

It depended on the link. That 900mhz bh I *****ed about for a year doubled in throughput when I increased bytes to 65000 on the AP side and lowered bytes to 3839 on the sta. I kept frames @ 32. The reason I lowered the sta side is because I sell asym access and most of the outbound from subs should be acks.

I lowered the bytes to 32000 on APs with subs in nLOS or NLOS areas and set all subs to 3839.

Still fooling with combinations...spent the last 2 days doing service calls because of dish spins. The pic shows how windy it gets around here.

[Apogee]

Aggregate size is adjusted depending on link conditions and what is specified in UI is the maximum. Lower priority CPEs get less slots but aggregation is not dependent on priority but on link conditions.

-Edmundas

This does not jive with my results so far.

Does am pri affect link conditions?