Does anyone have a good resource for acceptable speeds for home DSL customers?
I would like to see acceptable speeds from the customer CPE to the first layer 3 hop, the hop to the upstream and the hop that leaves the upstream network.
Thanks
luke
Luke Parrish
Centurytel Internet Operations
318-330-6661
Surely this is completely subjective?
Wearing my end-user hat, I see, and expect, TCP traffic flowing at
about 55kB/s on a BT Home 500 circuit, and proportionally higher
throughputs on the 1000 and 2000 offerings.
If it routinely fell below about 75% of the theoretical maximum even
to close sites, I'd switch providers.
My email was confusing since I said the word speed, I would like to ms roundtrip for the following:
1. CPE to first layer 3 hop
2. CPE to first layer 3 upstream hop
3. CPE to layer 3 exit point of upstream
Example:
Trace route to www.yahoo.com
Those ms values are what I am curious about. What are other providers seeing and what are, in your opinion, acceptable ms times for a home 1.5M dsl user…
Luke
I agree with that regarding download speed, I like to see 85-90% on a 1.5 connection, that usually works out to 1.3M - 1.35M...
And, yes this is purely for conversation and other viewpoints.
Thanks,
luke
I have found that "acceptable speeds" for residential users will vary widely from one area of the country to another. To a large degree it is a perception issue rather than an empirical one (ie www.cnn.com loads "too slowly"). The best metric for the happiness of a DSL customer base seems to be simply how many complaints you get and how many switch to cable modem. Granted, there are always a silent majority who are unhappy and will never let you know until they cancel, but the number of complaints you do get can usually be used to extrapolate the rest.
That being said, I think it would be a useful thing for a provider to have a local way to measure speed from the customer to some relatively close point in the network, and then you as a company can evaluate if your upstreams suckiness more accurately than a customer could. I think it would be reasonable to expect that a customer should get near line rate across your network and to the first hop of your upstream. After that, it depends on the suckiness factor.
Luke Parrish wrote:
Traceroute is not an effective measurement of performance. Due to the way routing devices process the packets it receives, it is possible for the latency that appears in a traceroute is far higher than the latency of traffic traversing that device.
Luke Parrish wrote:
Luke Parrish wrote:
My email was confusing since I said the word speed, I would like to ms roundtrip for the following:
*1. CPE to first layer 3 hop
2. CPE to first layer 3 upstream hop
3. CPE to layer 3 exit point of upstream*Example:
Trace route to www.yahoo.com
<http://www.yahoo.com/>1\. 10.10.10.1 (CPE) 1ms
2. 10.10.10.254 (DSLAM)(cte) 21ms*(first layer 3 hop)
*3. 11.1.1.1 (Router)(cte) 24ms
4. 5.5.1.3 (upstream interface)(level3) 68ms*(first layer 3 upstream hop)
*5. 5.4.3.2 (exit point of upstream)(handoff from level3 to at&t) 94ms *(layer 3 exit point of upstream)*Those ms values are what I am curious about. What are other providers seeing and what are, in your opinion, acceptable ms times for a home 1.5M dsl user...
Luke
The speeds will vary based on the packages built out. When using interleaved mode (for ADSL anyways), you will see somewhere along the lines of 20-30ms from the CPE to the DSLAM. When not using interleaved, I have seen 5-10ms between the CPE and DSLAM. Ofcourse, cable distances play into this as well I'm sure. And different technologies (SHDSL) will have different latency figures.
Yes, but I have to hold my upstream accountable for the level of service they provide to me and eventually to my end customer.
We have ways to measure download speed and ms response time from my network down to the customer and them from my network out to the internet via our upstream. However I am looking for benchmarks to compare these times against...
Andrew, traceroute is an effective tool is measuring roundtrip in ms format. I am not looking for download speeds or standards, I have already established those. Yes I agree, traceroute is not an effective tool for measuring download speeds.
thanks,
luke
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Luke Parrish wrote:
My email was confusing since I said the word speed, I would like to ms
roundtrip for the following:*1. CPE to first layer 3 hop
2. CPE to first layer 3 upstream hop
3. CPE to layer 3 exit point of upstream*Example:
Trace route to www.yahoo.com
<http://www.yahoo.com/>1\. 10.10.10.1 (CPE) 1ms
2. 10.10.10.254 (DSLAM)(cte) 21ms*(first layer 3 hop)
*3. 11.1.1.1 (Router)(cte) 24ms
4. 5.5.1.3 (upstream interface)(level3) 68ms*(first layer 3 upstream hop)
*5. 5.4.3.2 (exit point of upstream)(handoff from level3 to at&t) 94ms
*(layer 3 exit point of upstream)*Those ms values are what I am curious about. What are other providers
seeing and what are, in your opinion, acceptable ms times for a home
1.5M dsl user...
Those times seem high to me. I have a 1.5/768 ADSL circuit and I routinely
see 13-15ms to my 1st IP hop and 15-18 to the upstream handoff. I'm
14.5Kft from my CO and my IP is backhauled to SFO from SJC. Here are a few
examples:
bep@son-of-spike% traceroute www.yahoo.com
traceroute to www.yahoo.akadns.net (66.94.230.44), 30 hops max, 40 byte packets
~ 1 cerberus-internal.pinskyfamily.org (172.16.77.1) 3 ms 1 ms 1 ms
~ 2 er1.sfo1.speakeasy.net (66.92.1.1) 18 ms 17 ms 17 ms
~ 3 220.ge-0-1-0.cr2.sfo1.speakeasy.net (69.17.83.177) 16 ms 15 ms 15 ms
~ 4 bas1-m.pao.yahoo.com (198.32.176.135) 16 ms 15 ms 14 ms
~ 5 ge-1-0-2.msr1.scd.yahoo.com (66.218.82.193) 17 ms 15 ms 17 ms
~ 6 UNKNOWN-66-218-82-230.yahoo.com (66.218.82.230) 16 ms
vl42.bas1-m.scd.yahoo.com (66.218.82.226) 16 ms 16 ms
~ 7 p13.www.scd.yahoo.com (66.94.230.44) 18 ms 16 ms 15 ms
bep@son-of-spike% traceroute www.nytimes.com
traceroute to www.nytimes.com (199.239.137.245), 30 hops max, 40 byte packets
~ 1 cerberus-internal.pinskyfamily.org (172.16.77.1) 2 ms 1 ms 1 ms
~ 2 er1.sfo1.speakeasy.net (66.92.1.1) 19 ms 16 ms 16 ms
~ 3 110.ge-0-0-0.cr1.sfo1.speakeasy.net (69.17.83.189) 17 ms 14 ms 33 ms
~ 4 g8-1.mpr2.pao1.us.above.net (209.249.11.177) 14 ms 16 ms 14 ms
~ 5 so-0-0-0.mpr4.pao1.us.above.net (64.125.27.82) 13 ms 13 ms 13 ms
~ 6 p4-2-0-0.r06.plalca01.us.bb.verio.net (129.250.9.129) 15 ms 13 ms 18 ms
~ 7 p16-0-1-0.r21.plalca01.us.bb.verio.net (129.250.3.82) 15 ms 17 ms 16 ms
......
bep@son-of-spike% traceroute www.cisco.com
traceroute to www.cisco.com (198.133.219.25), 30 hops max, 40 byte packets
~ 1 cerberus-internal.pinskyfamily.org (172.16.77.1) 3 ms 2 ms 1 ms
~ 2 er1.sfo1.speakeasy.net (66.92.1.1) 14 ms 14 ms 15 ms
~ 3 120.ge-0-0-0.cr2.sfo1.speakeasy.net (69.17.83.185) 13 ms 50 ms 14 ms
~ 4 ge-4-0-440.ipcolo1.SanJose1.Level3.net (209.247.156.221) 13 ms 12 ms
~ 13 ms
~ 5 p1-0.cisco.bbnplanet.net (4.0.26.14) 19 ms 13 ms 14 ms
~ 6 sjce-dmzbb-gw1.cisco.com (128.107.239.53) 13 ms 13 ms 13 ms
~ 7 sjck-dmzdc-gw1.cisco.com (128.107.224.69) 15 ms 13 ms 13 ms
......
bep@son-of-spike% traceroute www.cnn.com
traceroute to cnn.com (64.236.24.12), 30 hops max, 40 byte packets
~ 1 cerberus-internal.pinskyfamily.org (172.16.77.1) 2 ms 1 ms 1 ms
~ 2 er1.sfo1.speakeasy.net (66.92.1.1) 19 ms 15 ms 15 ms
~ 3 210.ge-0-1-0.cr1.sfo1.speakeasy.net (69.17.83.181) 17 ms 15 ms 13 ms
~ 4 g8-1.mpr2.pao1.us.above.net (209.249.11.177) 17 ms 14 ms 14 ms
~ 5 so-4-2-0.mpr3.sjc2.us.above.net (64.125.28.222) 17 ms 16 ms 16 ms
~ 6 so-0-0-0.mpr4.sjc2.us.above.net (64.125.30.2) 18 ms 16 ms 16 ms
~ 7 so-3-3-0.cr1.dfw2.us.above.net (64.125.29.58) 64 ms 61 ms 61 ms
~ 8 so-4-0-0.mpr1.iah1.us.above.net (64.125.31.37) 65 ms 66 ms 75 ms
~ 9 so-0-0-0.mpr2.iah1.us.above.net (64.125.31.62) 65 ms 66 ms 65 ms
10 so-5-0-0.mpr1.atl6.us.above.net (64.125.29.65) 77 ms 77 ms 75 ms
11 aol-above.atl4.above.net (209.249.119.242) 75 ms 85 ms 79 ms
12 bb1-atm-P0-0.atdn.net (66.185.147.192) 76 ms 76 ms 75 ms
13 pop1-atl-P4-0.atdn.net (66.185.136.17) 74 ms 75 ms 75 ms
...........
- --
Expect ~20% less than rated speed for ATM overhead.
Expect 20-40 ms on first hop due to DSLAM interweaving.
James H. Edwards
Routing and Security Administrator
At the Santa Fe Office: Internet at Cyber Mesa
jamesh@cybermesa.com noc@cybermesa.com
http://www.cybermesa.com/ContactCM
(505) 795-7101
* Luke Parrish:
Andrew, traceroute is an effective tool is measuring roundtrip in ms
format.
No, it's not, because routers generate ICMP TTL Exceeded packets with
totally different machinery, separated from the forwarding path. Many
factors influence the ms numbers traceroute reports (MPLS, main CPU
load, priority on internal busses), and only some of them correlate
with forwarding latency.
Andrew, traceroute is an effective tool is measuring roundtrip in ms format.
packets sent to a router are typically processed differently and with different priority then packets forwarded through it. This makes traceroute fairly unreliable.
I measure acceptable performance on the basis of the needs of my applications. In general I need <= 100ms rtt and little or no re-ordering to the counterstrike hl2 servers I play on. providers that can't deliver that reliably don't get my business.
Joel Jaeggli wrote:
Andrew, traceroute is an effective tool is measuring roundtrip in ms format.
packets sent to a router are typically processed differently and with different priority then packets forwarded through it. This makes traceroute fairly unreliable.
Since it is probably a fair assumption that routers will never procces forwarding packets slower than ICMP replies, the following applies.
The router receiving the traceroute response from its upstream would process that in its forwarding path. So if you see a 30ms hit on hop A and a 60 ms hit on hop B you can pretty much determine that hop A is 30ms away but you cant be quite sure about hop B until you see hop C's replies.
To make this more interesting, its always possible that hop B or C's path to you is different than your path to hop B or C.
Also, traceroute is effective at showing that the path rtt is good. Its just when you are trying to find where the latency is that things can get dicey.
Try:
http://www.dslreports.com/archive?all=1
to see how you compare with others in your ISP or area (you can search by
zip code).
Regards,
Hank
I think we are WAY overanalyzing...
The only tool a typical end user has for testing is ping and tracert on their windows machine. They do not care about packet processing, icmp priority, processor speeds, forwarding paths, mpls, internal bus, they only care about what the ms latency is between each of their trace route hops. They do not care if it is accurate or is even relevant. When they call into support they only care that their buddy down the road gets 45ms to yahoo.com with BST and gets 82ms with Centurytel.
I have always noticed that about this list which is why I rarely post. Seems that people spend more time picking apart your question then actually answering it.
I am not asking which tools are best or most accurate. I am not asking if ms latency is an accurate measure of DSL service. I am not asking about DSLAM configurations. I am simply asking what a typical and acceptable traceroute/ms latency is for a home DSL account.
I have received several very good answers that have answered my questions perfectly. Thank you to all that answered. I have enough information to accomplish my end goal now, again thanks for the help.
Luke
When I switched from 1600/384 to 3000/768 dsl, download speed went up to
very nearly the promised 3Mbps, but latency to the first hop went from
14 ms to 26 ms.
Now I have FTTH, and first-hop latency is 3 ms (acedsl.com, Verizon
reseller, good guys).
I assume it had to do with different settings for interleaving on the
DSLAM, as some prior poster mentioned.
>
> >When I switched from 1600/384 to 3000/768 dsl, download speed went up
to
> >very nearly the promised 3Mbps, but latency to the first hop went from
> >14 ms to 26 ms.
>
> Is there a reason for that ? that, latency goes up when bandwidth goes
up
> for your case ?
I assume it had to do with different settings for interleaving on the
DSLAM, as some prior poster mentioned.--
Barney Wolff http://www.databus.com/bwresume.pdf
I never met a computer I didn't like.
Interleaved adds some error correction, allowing the connection to be more
resistant to interference (noisy lines), but at the expense of latency.
Fast-path is the other way data is sent, which obviously, is much faster,
I've personally seen less ~10ms for a loop around 11000ft.
I think we are WAY overanalyzing...
Oh, we're just getting started. 
The overanalyzing is a result of the fact that in a *very* large percent of
cases, the question that's asked isn't the question that the asker really wants
answered. Even simple "Where did they hide the router config option that does
XYZ?" questions often take a very sharp left turn into "They intentionally hid
it because you probably don't actually want that, as this other thing is
usually a better solution".
And especially on a list like this one, where the list participants are busy
trying to sell into wildly different value propositions, there are often
multiple correct answers, and they are often *very* sensitive to boundary
conditions. The set of answers that works for my employer probably won't
work for the average cablemodem ISP, and neither of our answers will be right
for people who are trying to buy/sell an OC-192 worth of transit to each other.
For that matter - our answers often aren't interchangeable with another
public university just 10 miles down the road (in fact, they *were* a part
of us from 1944 to 1964) simply because we're 4 times bigger and they're
liberal arts oriented.
I have always noticed that about this list which is why I rarely post.
Seems that people spend more time picking apart your question then actually
answering it.
Well... the *original* question was "What's an acceptable speed for DSL?", and
the only *really* correct answer is "The one that maximizes your profit
margin", balancing how much you need to build out to improve things against
whatever perceived sluggishness ends up making your customers go elsewhere. As
noted elsewhere, it really depends on the hoover quotient of the other DSL and
cablemodem providers with a presence in the area.
The average user may have ping and tracert, but couldn't figure out how to use
them or interpret the results even if you stapled a cheat sheet to their
forehead. It's strictly "www.cnn.com is/isn't acting piggy". If you find a
user who figures out how to open a CLI window and launch ping, you have a
*geek* user on your hands - and at *that* point it's of course totally
acceptable to go into turbo-geek mode and discuss the forwarding paths inside
the routers.