PMP320 ping times above 600

We have a small to medium 320 network. We have been running it for a while now. When we first set it up, everything worked very well.

Now we are seeing clients with really high ping times. When they do a speed test, it comes out with what their QOS setting is set for, but the pings are well over 500 and some are over 600.

We are using later2 on the AP's and bridge mode on the SM's. We are not using HARQ in the QOS. We static all the SM's and we static all the client routers.

We run licensed backhauls in 11ghz. Not sure on what to check??? Any help would be great. Thanks!!!

Where I'd look first if it were me:

Check how much AP traffic there is (I assume you're graphing the traffic somewhere?).  If most of your subs aren't MIMO-B or have weaker signals/modulations, the AP may be maxing out it's available throughput at a lower value than you might have expected.

Try running the Frame Utilization tool (Performance Monitor->Frame Utilization) and see if your downlink or uplink slots seem to be at or near 100% capacity during these high ping times.

If the RF frame looks saturated in the above tool, check what the modulation levels actually being passed are from your subs (Active Service Flow Stats->Downlink MCS and Active Service Flow Stats->Uplink MCS), clear stats - let some data pass and watch for people passing most/all of their traffic at QPSK modulations - if you have a few of these poor signal subs slamming your AP at QPSK modulations, try to send service calls to improve their RF signals first.

If you're not at 10MHz channels, consider whether you can move toward a 10MHz channel for improved capacity (pending subscriber signals being strong enough and there being adequate SNR).

Check CINR levels on both uplink and downlink - do you suspect any interference?

Odd.  I swear I responded to this a few days ago.  Maybe it didn't like the spreadsheets I put up to help with number crunching.

There's a lot of useful info in the 320s, the problem if figuring out what to do with it.

Keep in mind that I'm in a very hilly, heavily treed, and noisy environment, so my suggestions are all based on absolute worst case scenario.

I would strongly suggest using HARQ on both UL and DL, even if only with a single retry.  That gives you the HARQ stats, which I found very useful as an indication of stability.  If you're seeing a lot of NAKs, then your best bet is to increase HARQ retries and your C/N backoffs.  You want to strike a balance between the two - 1 retry at a low modulation will be slower overall than 5 retries at a higher modulation and vice versa.  The active service flow stats are helpful in terms of seeing discards, but I found the HARQ stats more useful.

The MCS stats for up and down are also very useful.  If you're willing to do some number crunching, you can use them to combine bandwidth usage and modulation stats to see which CPEs are using the most frame slots.  If an AP is saturating, that will show you who needs to be throttled back or realigned.

A note about frame utilization - you want to turn both the number of frames and the time between analyses up quite a bit.  The default is too small a sample size so it gives very unpredictable results, and if you max it without turning up the delay the AP will end up analysing every single frame and taking a performance hit which could invalidate the test.

One last thing - I found the CINR Reuse 1 was the magic number.  It was consistently the lowest of the CINR values, and the lowest value is what determines what modulation a customer ends up using.  In some cases, I had very nice looking CINR values and RSSI, but low reuse 1 numbers, and the customer was locked to QPSK as a result.