[LARTC] Slow large pings

[LARTC] Slow large pings

[kasp]

Hello lartc,

Testing some rules I found out that any rules are slowing down large pings!
For example:

tc qdisc add dev eth0 root handle 1: htb
tc class add dev eth0 parent 1: classid 1:1 htb rate 100Mbit prio 1 ceil 100Mbit
tc class add dev eth0 parent 1:1 classid 1:2 htb rate 100Mbit prio 2 ceil 100Mbit
tc filter add dev eth0 parent 1:0 prio 5 u32 \
    match ip src 10.10.10.1 \
    match ip dst 10.10.10.10 \
    flowid 1:2
tc qdisc add dev eth0 parent 1:2 handle 2: sfq perturb 10

Small pings are ok - <10ms, here is output for 65500 byte pings:

Pinging test [10.10.10.1] with 65500 bytes of data:

Reply from 10.10.10.1: bytese500 time\x120ms TTLd
Reply from 10.10.10.1: bytese500 time\x120ms TTLd
Reply from 10.10.10.1: bytese500 time\x120ms TTLd
Reply from 10.10.10.1: bytese500 time\x120ms TTLd

Ping statistics for 10.10.10.1:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 120ms, Maximum =  120ms, Average =  120ms

If I don't use any rules, 65500 byte pings are about ~10ms
Any filters are slowing large pings for 10 times!!!
What can I do to tune it up?

-- 
Best regards,
 kasp                          mailto:waters@inbox.lv

_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

Re: [LARTC] Slow large pings

[Andy Furniss]

kasp wrote:
> Hello lartc,
> 
> Testing some rules I found out that any rules are slowing down large pings!

I can't reproduce this.

> For example:
> 
> tc qdisc add dev eth0 root handle 1: htb
> tc class add dev eth0 parent 1: classid 1:1 htb rate 100Mbit prio 1 ceil 100Mbit
> tc class add dev eth0 parent 1:1 classid 1:2 htb rate 100Mbit prio 2 ceil 100Mbit

> tc filter add dev eth0 parent 1:0 prio 5 u32 \

Error for me unless I add protocol ip

>     match ip src 10.10.10.1 \
>     match ip dst 10.10.10.10 \
>     flowid 1:2
> tc qdisc add dev eth0 parent 1:2 handle 2: sfq perturb 10
> 
> Small pings are ok - <10ms, here is output for 65500 byte pings:

10ms is high I get 0.3 with p200mmx +rtl8139 + cheapest switch I could 
find + cables made without crimping tool.

> 
> Pinging test [10.10.10.1] with 65500 bytes of data:
> 
> Reply from 10.10.10.1: bytese500 time\x120ms TTLd
> Reply from 10.10.10.1: bytese500 time\x120ms TTLd
> Reply from 10.10.10.1: bytese500 time\x120ms TTLd
> Reply from 10.10.10.1: bytese500 time\x120ms TTLd
> 

I see 14ms

> Ping statistics for 10.10.10.1:
>     Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
> Approximate round trip times in milli-seconds:
>     Minimum = 120ms, Maximum =  120ms, Average =  120ms
> 
> If I don't use any rules, 65500 byte pings are about ~10ms
> Any filters are slowing large pings for 10 times!!!
> What can I do to tune it up?
> 

What kernel and hardware are you using?


Andy.

_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/