Long 'ping' time? 3

[lamsf]

What is the reason behind very long 'ping' time? Tq.

 

[Grenage]

Latency?

Carlsberg don't run I.T. departments, but if they did they'd probably be more fun.

 

[jimbopalmer]

We do not have many details about your setup: smoke signals, dial-up modem, satellite, ISDN, DSL, T1, Cable Modem, T3, OC192, etc. We may assume you have a computer of some sort, but we don't know if you have a SOHO router, or an entire University connecting you to your destination, which might even involve the Internet, we don't know.

If you computing device supports it, I would recommend traceroute over ping, it gives way more detail we could use to offer real help.

If everything else is perfect, the speed of light through fibre limits your ping.

http://www.stuartcheshire.org/rants/Latency.html

I tried to remain child-like, all I acheived was childish.

 

[lamsf]

Hi, very sorry for the vague question. I will try to explain it here.

We did a continuous ping. Normally, the latency is about 80-120 ms. However, at times it will reach upto >300ms. Any reason what contributes to the high latency?

Actually, will ping encounter retransmission? Or, it will just wait till it timeout?

Thank you.

 

[jimbopalmer]

Some how you missed out telling us ANY of the variables mentioned: What computer? What OS? What network technology? What access method? What ISP? We did not convince you to show us a traceroute, we could speculate all week, but YOU have the ability to make this go quickly if YOU choose to tell us.

Each ping either gets to it's destination or not. The destination either replies or not. And the reply either gets back to us, or not.

I tried to remain child-like, all I acheived was childish.

 

[lamsf]

Very sorry. I didn't know those information is so important. I will try to explain more.

We are setting up a wifi network. We have a laptop (running on windows) connected to the wifi network and we make a continuous ping to one of the servers (running on windows) connected to the core network of the wifi network. So, there is no ISP involved here.

When we did the continous ping, the normal latency is between 80-120ms. However, at times, it will reach upto >300ms. We would like to know what could be the reason.

I cannot do the trace at the moment. Will try doing it later. Without the information from trace, will you be able to help?

Thank you.

 

[lamsf]

Hi, another question.

Is there any retransmission for ICMP or ping packets?

Tq.

 

[Grenage]

Only when they time out and the computer pings again. You'll want to plug the laptop directly into the router, to see if the ping time comes down. If it does, then it might be worth looking at possible wifi interference or a defecting AP.

Carlsberg don't run I.T. departments, but if they did they'd probably be more fun.

 

[lamsf]

Only when they time out and the computer pings again. You'll want to plug the laptop directly into the router, to see if the ping time comes down. If it does, then it might be worth looking at possible wifi interference or a defecting AP.

Thanks.
So, it won't automatically retransmit?

For the continuous ping test that we did, most of the time the latency will be within the range of 80-120ms. Only occassionally there will be a spike, upto >300ms. For eg. out of 100 pings, maybe 5 pings will hv latency >300ms.

Is there anyway to investigate why that few pings exceed latency of >300ms?

Thank you.

 

[jimbopalmer]

The 'ping' program will only send one packet per ping.
At an IP layer, checksums are still being checked for corruption, so it is possible for a ping packet to generate a ICMP error reply.

Normally the physical layer has little error correction in it, but Wireless trades really bad wire (air) for quite a lot of electronics. You would have to read quite a lot of WiFi specs to know if retransmits could be happening in the radio network than never show in the data network.

I still feel we are not getting the whole picture, 80 ms pings ought to be able to connect the east and west coasts of the US, not computers across a room.

I tried to remain child-like, all I acheived was childish.

 

[jimbopalmer]

My wife's wireless XP laptop at the front of the house (the only wireless device we have) gets as low as 1 ms pings, as high as 1500 ms pings, and averages 17 ms over 10 minutes.

I tried to remain child-like, all I acheived was childish.

 

[burtsbees]

To all:
Wireless transmissions that are full duplex (802.11a/b and g) have extra error correction and acknowledgement overhead. You will see an initial greater period of latency versus wired transmissions.

Burt

 

[jimbopalmer]

Burtsbees I do not believe all of the 802.11 variations are full duplex. The original 802.11 and 802.11b both use CSMA/CA.

802.11a and 802.11g use TDM, so may approach full duplex, but 802.11g still suffers badly if multiple clients are online. (in wired networks, full duplex implies clients that do not conflict, which may not apply to WiFi)

I tried to remain child-like, all I acheived was childish.

 

[burtsbees]

Oh...I was trying to think back to what my instructor in the Cisco Academy mentioned about that a few yers ago...sorry. I do remember him telling me that there was a lot of extra overhead with wireless, like 54MBps is actually not what you get, or something like that. Thanks for the info.

Burt

 

[lamsf]

If I would like to investigate the high latency on ping, what is the best way to do it? Is there any TCP/IP analysing tool which can do that? Thanks.

 

[jimbopalmer]

Ping is one of TCP/IP's latency analyzing tools, the other is traceroute.

Sadly, no matter how we ask, you refuse to give us enough information to help you do any analysis.

I tried to remain child-like, all I acheived was childish.

 

[burtsbees]

That is likely the nature of wireless. You did not get the latency before WiFi, right???

Burt

 

[lamsf]

I hv made some pings and tracert:

Reply from 172.16.10.22: bytes=100 time=41ms TTL=126
Reply from 172.16.10.22: bytes=100 time=81ms TTL=126
Reply from 172.16.10.22: bytes=100 time=86ms TTL=126
Reply from 172.16.10.22: bytes=100 time=81ms TTL=126
Ping statistics for 172.16.10.22:
Packets: Sent = 50, Received = 50, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 41ms, Maximum = 233ms, Average = 77ms


Tracing route to 172.16.10.22 over a maximum of 30 hops
1 <1 ms <1 ms <1 ms 192.168.15.1
2 * * * Request timed out.
3 63 ms 74 ms 74 ms 158.171.52.129
4 88 ms 69 ms 69 ms 172.16.10.22
Trace complete.

Do you think there is any problem?

If there is a surge on the round trip times (eg. 233ms), is it possible that we find out the reason for the spike? It could not be due to any retransmission, right?

Tq.

 

[jimbopalmer]

Tracing route to 172.16.10.22 over a maximum of 30 hops
1 <1 ms <1 ms <1 ms 192.168.15.1
2 * * * Request timed out.
3 63 ms 74 ms 74 ms 158.171.52.129
4 88 ms 69 ms 69 ms 172.16.10.22
Trace complete.

OK, this is wild eyed guesses, be cause I do not know anything about these devices, but:

Device 1 I am betting 192.168.15.1 is your Wireless router, if so, you have less than 1 ms to it which is great, and rules out this being a wifi issue.

Device 2 is configured not to respond to ICMP packets, which makes it a challenge to trouble shoot.

Device 3 158.171.52.129 is owned by the Bank of Boston, if that is not you, I worry that you may be 'losing' part of the internet by using public IP addresses for private uses.

If this link is configured correctly, it would seem to be at least 5000 miles round trip from Device 1. With Device 2 not responding, it is hard to guess how much of the mileage is between 2 and 3 (perhaps a traceroute in the other direction would help)

Device 4 is nearby to Device 3 and is not the issue.


I tried to remain child-like, all I acheived was childish.

 

[LawnBoy]

Welcome to wireless. You'll never achieve the latency and stability of a wired network. Remember, it's just a radio signal, and very susceptible to external noise.