What’s the general best idea for framer policy on an n-streme ptmp link?
I’ve been doing best fit but don’t have any idea of it’s merits vs.
the other options.
We are in the same situation. We have tried various settings, but fail to notice any difference running bw tests. best fit, exact-size.. what size?. Anyone have suggestions or can refer to related literature?
Anyone have any input on this?
ATM i’m using dynamic size 3200
Exact size frames the packets to match the exact size you entered. This makes higher bw output, but many data loss may occurr: if a packet is framed in 2 [exact size] and one [exact size] is lost, this packet has to be resent aswell and it may increase latency.
Best fit doesn’t frame any packet, but you would experience a bw loss.
Dynamic size doesn’t frame any packet aswell and it dinamically resize the nstreme packet. It may increase CPU usage, since the radio has to calculate the size of every nstreme packet.
Please Normis correct me if i’m wrong.
Hello from Greece,
Do you have any input on this?
What do you recommend to use?
I tested this about a year ago in a clean PtP configuration and found that best fit and 3200 where the best for increasing throughput.
Of course this assumes that you are running CCQ close to 100%. The more noise you have, the less improvement there is with nstream. At some point, nstream makes it worse.
So the question in PtMP is, how much noise are you dealing with? If there is a lot, then keep the packet sizes small or you will lose more data with nstream.
Tom