I tested ping times when I do an upload speed test. The result is not pretty.
Test 1) Ping on RG no other activity
--- 206.190.36.45 ping statistics ---
18 packets transmitted, 18 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 45.294/46.189/47.486/0.681 ms
Ping time are good, little variation.
Test 2) Ping on RG at the same time as a speed test
--- 206.190.36.45 ping statistics ---
41 packets transmitted, 40 packets received, 2.4% packet loss
round-trip min/avg/max/stddev = 45.112/244.032/1788.340/387.783 ms
Speed Down/Up 15.14Mbs/1.67Mbs
Ping times all over the place, we now see how a ping is affected by other activity especially uploads. This is very bad caused by big buffer on the RG and slow upload speed.
Test 3) Ping on Cerowrt router in DMZ no other activity
--- ds-any-fp3-real.wa1.b.yahoo.com ping statistics ---
30 packets transmitted, 30 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 45.461/45.961/48.741/0.580 ms
Ping time are good little variation. Test to get a base line and see that nothing is wrong.
Test 4) Ping on Cerowrt router in DMZ at the same time as speed test no QOS/SQM
--- 206.190.36.45 ping statistics ---
52 packets transmitted, 52 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 45.348/158.877/1190.716/289.484 ms
Speed Down/Up 15.30Mbs/1.79Mbs
Ping times allover the place, a little better than test 2 probably because the Cerowrt router has better wifi buffer management. This is still very bad.
Test 5) Ping on Cerowrt router in DMZ at the same time as speed test with QOS/SQM enabled
--- 206.190.36.45 ping statistics ---
55 packets transmitted, 55 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 45.192/46.824/51.597/1.481 ms
Speed Down/Up 14.41Mbs/1.36Mbs
Ping time are good little variation. This is nice the QOS or SQM(Smart Queue Management as the Cerowrt calls it) limits upload to less than line speed so the RG upload buffer is always empty. Speed drops a little but I get a more responsive broadband which is what I want.
Takeaway
1) Measuring only latency or only speed is not good enough you need to test both at the same time.
2) RG are not good devices, equipment manufactures need to fix their buffer management.
3) Cerowrt does it right and can even compensate for the RG problems. Cerowrt give back their stable changes to Openwrt. I do not have Openwrt to test.
I have 12M/1.5 uverse service. The RG is a Pace 5031NV. All test where done over wifi. Speed test was Ookla. Cerowrt router is a Netgear WNDR3800 with build 3.10.50-1.
↧