Testing Latency

[BarkingGhostar]

This week I tested two scenarios in my home:

Scenario A: Comcast 25x5 Mbps connection coming into home through a Motorola cable modem that goes into a Linksys router with wifi turned off. The router's LAN ports feed several sections of my home and at theis 'A' testing point comes into an old Linksys WRT54G (Yeah Baby!) that is configured as a wireless switch. I use a LAN port on the back to feed into a W7-64/16GB RAM/i7-4770K. NIC is configured for 100Mbps Full Duplex.

Scenario B: AT&T Fiber 1x1 Gbps connection coming into my home through a Pace 5628 gateway that goes directly into a Dell laptop W7-64/8GB RAM/i7 4770M? NIC is configured for Gbit Full Duplex. Skirts the household network for now as it is in a pure testing phase.

'A' yields 30x6 Mbps speed and latency of 11ms. It is consistent and has been consistent for the most part. The latency actually got worse a couple of years ago as it was 8 ms.

'B' yields 0.934x0.904 Gbps speed and a latency of 29-32 ms.

I can get these results from a variety of testing websites and also latency performance ratios between the two services remains even when doing simple ICMP testing from a DOS prompt. And while I would imagine my home network would introduce more latency (add to it) the worse latency performer is actually skirting the home network and the better perform goes through the home network.

So, what am I missing here (if anything)?

 

[Fardringle]

Latency can vary significantly depending on the route that the data takes to get to the destination, and different ISPs frequently use very different routes.

That being said, 11ms vs 32ms on an Internet connection is insignificant unless you are running applications that depend on a specific amount of latency. You'll never notice it in web browsing, video streaming, games, downloads, or anything else that would normally be running on a home Internet connection.

 

[JackMDS]

The first info on good tracer is the latnecy Between the Computer and the local Router. (usually it is 0 msec.).

From there On it is the Internet Routes One can do nothing about it short of Wishful Thinking and Dreaming. -

 

[BarkingGhostar]

Jack, the scenario with the worse latency is directly connected to the AT&T residential gateway and not going through any of the home network, whereas the better latency scenario is going through several hops before exiting the home. I interpreted this as the latency observation being more of an external factor (not in my home) than internal.

 

[Fardringle]

I interpreted this as the latency observation being more of an external factor (not in my home) than internal.

Yep. If you ping your router directly (ping the Default Gateway address) and get anything higher than 1ms, then there's an internal problem. Otherwise it's on the ISP's network or beyond their network and you can't do anything about that.

 

[JackMDS]

Use this free portable App to Trace.

http://www.softwareok.com/?seite=Microsoft/TraceRouteOK

In the example bellow the trace is from my desktop to Microsoft.com

The first line is the timing from the Desktop to the Router, has you see it is 0msec.

From there on it is all traces that are beyond my control.


Host-Name: www.microsoft.com

1 / 0 ms | 0 ms | 0 ms 192.168.1.254 router.asus.com

2 / 12 ms | 13 ms | 15 ms 142.254.161.37 Unbekannt / Timeout 1000 Msek. (IP:142.254.161.37 )
3 / 8 ms | 10 ms | 13 ms 68.173.192.29 Unbekannt / Timeout 1000 Msek. (IP:68.173.192.29 )
4 / 10 ms | 13 ms | 17 ms 68.173.198.18 agg112.nyquny9101r.nyc.rr.com
5 / 14 ms | 16 ms | 22 ms 107.14.19.22 bu-ether25.nycmny837aw-bcr00.tbone.rr.com
6 / 8 ms | 8 ms | 9 ms 205.197.232.13 Unbekannt / Timeout 1000 Msek. (IP:205.197.232.13 )
7 / 7 ms | 9 ms | 12 ms 207.88.14.150 207.88.14.150.ptr.us.xo.net
8 / 10 ms | 12 ms | 15 ms 216.156.16.133 216.156.16.133.ptr.us.xo.net
9 / 8 ms | 9 ms | 11 ms 209.220.16.170 209.220.16.170.ptr.us.xo.net
10 / 12 ms | 12 ms | 13 ms 104.107.49.240 a104-107-49-240.deploy.static.akamaitechnologies.com

 

[dave_the_nerd]

Your home network, given that it's all wired, is adding <1ms of latency.

Comcast hosts dedicated speed test servers for a lot of those web benchmarking sites. It ensures that their customers see significantly better speed/latency numbers than they really have any right to. They also do some co-location for Netflix and a few other companies too.

What you should probably be doing is comparing latency/ping and traceroutes to specific frequently used sites. (Facebook, Google, your work's VPN connection, your favorite World of Warcraft server, etc.)

30ms is plenty fast, though, in any case.

 

[BarkingGhostar]

I understand I have no control over the network outside my home. I was making an observation and then surprised at what I was observing. I figured the direct connection would be with less latency and when it wasn't it surprised me, and maybe even a bit disappointing. However, last night I wanted to try to see if my typical hobbies were noticeably impacted and so I stuck my Xbox One on the Pace 5628 residential gateway. Before i had actually done anything with it I also connected my wife's iMac.

Using DSLReports' ping/jitter test, I compared the latency reporting for both the AT&T and Comcast connections as described above and the Comcast continued to produce 11ms times but now the AT&T Fiber had dropped to 1.6ms. I then tried Amazon Prime and One COD game play with not noticeable impact.