I was doing some testing on new systems yesterday and noticed that the RTT for a ping with jumbo frames was around 2.7ms. (average) while the same ping with normal frames was around .7ms (average). Although the difference seems proportionate, the RTT seems excessively high. If 15ms. latency is the tolerance for good SAN performance, 2.7ms. is an alarming portion of that. I also logged-in to a number of other systems and tested against a number of other SANs and found strikingly similar results. Any idea or comment on long RTT?
[root@esxi2:~] vmkping -d -s 8972 -I vmk1 10.0.0.64
PING 10.0.0.64 (10.0.0.64): 8972 data bytes
8980 bytes from 10.0.0.64: icmp_seq=0 ttl=64 time=2.920 ms
8980 bytes from 10.0.0.64: icmp_seq=1 ttl=64 time=2.804 ms
8980 bytes from 10.0.0.64: icmp_seq=2 ttl=64 time=2.591 ms
--- 10.0.0.64 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 2.591/2.772/2.920 ms
[root@esxi2:~] vmkping -d -s 1472 -I vmk1 10.0.0.64
PING 10.0.0.64 (10.0.0.64): 1472 data bytes
1480 bytes from 10.0.0.64: icmp_seq=0 ttl=64 time=0.934 ms
1480 bytes from 10.0.0.64: icmp_seq=1 ttl=64 time=0.816 ms
1480 bytes from 10.0.0.64: icmp_seq=2 ttl=64 time=0.500 ms
--- 10.0.0.64 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 0.500/0.750/0.934 ms
[root@esxi2:~] vmkping -I vmk1 10.0.0.64
PING 10.0.0.64 (10.0.0.64): 56 data bytes
64 bytes from 10.0.0.64: icmp_seq=0 ttl=64 time=0.470 ms
64 bytes from 10.0.0.64: icmp_seq=1 ttl=64 time=0.618 ms
64 bytes from 10.0.0.64: icmp_seq=2 ttl=64 time=0.345 ms
--- 10.0.0.64 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 0.345/0.478/0.618 ms
[root@esxi2:~]
+The Invisible Admin+
If you find me useful, follow my blog: http://johnborhek.com/