Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
http://usage.sbbnet.com/customer.php?customer=173.195.176.6&curDate=1345758487 That page shows the usage by hour/day/month/year of our speedtest server. It takes a while to load, its on an old server. I get a lot of tests from Apple, Inc (we are their nearest speedtest site). These shows up to and over 100mbps sometimes. But they are only for a few seconds each. It averages out to about 7mbps of upload data needed. We have lots of spare upload. On 08/22/2012 05:26 PM, Blake Covarrubias wrote: Matt, How much bandwidth on average are you seeing to your speed test servers? We're considering becoming a speedtest.net host, but are concerned about the amount of bandwidth generated by users outside our network. -- Blake Covarrubias On Aug 22, 2012, at 9:54 AM, Matt Jenkins wrote: We setup a speedtest server. So when customers connect to speedtest.net they reach our local server. The test data never leaves our network. On 08/21/2012 12:27 PM, Joey Craig wrote: We only guarantee from the customers radio to our servers and we make sure the customer is aware of that. We use Jperf/Iperf also…works very well. How can you guarantee ANYTHING beyond your network? You are absolutely in the right. Joey Craig Network/RF Engineer Firenet1.Com Phone: (662) 510-0764 Mobile: (662) 404-1118 From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On Behalf Of Bret Clark Sent: Tuesday, August 21, 2012 2:22 PM To: WISPA General List Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
A very common occurring reason for what you are describing would be a 'Ethernet Connection / Speed / Duplex Mismatch', in the customer's connection path. Regards Faisal Imtiaz Snappy Internet & Telecom On 8/23/2012 9:52 AM, Bret Clark wrote: I initially brought up the question because I have Verizon FiOS at home and not matter what time of the day can run a speed test to the same server and see my 30Mbps DN/15 Mbps up. If I run and Iperf test between my FiOS connection and one of our wireless business customers with 10Mbps circuit...it nails the speed every time... iperf show 10Mbps up & 10Mbps down (we have a Linux server at the customers site). Yet when that same customer runs the same speed.net test to the same speed.net server the results suck pretty much every single time! I know the majority of test speed test are really based on algorithms that try to figure out what the potential speed of a circuit is and because our network is all wireless it just screws up the algorithms? Bret On 08/22/2012 01:00 PM, Matt Jenkins wrote: http://www.speedtest.net/mini.php On 08/21/2012 12:22 PM, Bret Clark wrote: We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
I initially brought up the question because I have Verizon FiOS at home and not matter what time of the day can run a speed test to the same server and see my 30Mbps DN/15 Mbps up. If I run and Iperf test between my FiOS connection and one of our wireless business customers with 10Mbps circuit...it nails the speed every time... iperf show 10Mbps up & 10Mbps down (we have a Linux server at the customers site). Yet when that same customer runs the same speed.net test to the same speed.net server the results suck pretty much every single time! I know the majority of test speed test are really based on algorithms that try to figure out what the potential speed of a circuit is and because our network is all wireless it just screws up the algorithms? Bret On 08/22/2012 01:00 PM, Matt Jenkins wrote: http://www.speedtest.net/mini.php On 08/21/2012 12:22 PM, Bret Clark wrote: We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
that is a common situation, but another one is: "Sir, you had to have some virus or p2p application..." "no, you changed something now it's working well, so you are not telling me all the truth, it's working good now only because I called to complain..." We did nothing but they keep not believe us. Sometimes I feel that the more you tell the truth the more they will not believe you Maybe I could have better success with "yes, you are right we catch all the packets coming from complaining users and we give them higher priority by hand, so the more you call us the faster you go..." hum... unfortunately they do not pay us to spend time at the phone or that could be a good approach :) > For us this is the biggest one. > I will have customer call, speedtest says 384K. Look at the radio and > we are pushing 2.04M in on a 2M connection. "Ask your kids to turn off > all their devices and try again." "Oh, wow, I am getting the full 2M > now. What did you change?" Then the long explanation about shared > bandwidth on their connection. > Has to be the same for businesses, though it is kind of hard to have all > the other kids turn off their computers. > > On 8/21/2012 3:33 PM, Larry Weidig wrote: >> >> Bret: >> >> We have found that the tests are “fairly” accurate (well speedtest.net >> anyhow), *_BUT_* only in controlled environments. Your end user is >> probably running these tests when their Internet is “slow”. Viola >> look I am right it is slow, xyz speed test says so! What he does not >> realize is that Johnny intern in the warehouse has set his computer up >> to grab torrents for him while he picks stock and then just dumps the >> off to flash before heading out. Ok, that is just one example, but >> all sorts of other traffic can come into play. >> >> We tell our customers that we are not at all interested in speed tests >> unless we have a support person on the line watching their link for >> other types of traffic as well during the test. We also host their >> speedtest.net mini (free) on own of our own servers so that we can >> eliminate any Internet oddness causing problems with the results. The >> mini however only seems accurate up to about 25-30 Mbps in our >> experience at which point the results are only ballpark at best. Ookla >> claims mini is good until about 50 Mbps, but that has not been our >> experience. >> >> Just what we have experienced with speedtest results. >> >> Larry A. Weidig (lwei...@excel.net <mailto:lwei...@excel.net>) >> >> Excel.Net, Inc. – http://www.excel.net/ >> >> (920) 452-0455 – Sheboygan/Plymouth area >> >> (888) 489-9995 – Other areas, toll-free >> >> *From:*wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] >> *On Behalf Of *Bret Clark >> *Sent:* Tuesday, August 21, 2012 2:22 PM >> *To:* WISPA General List >> *Subject:* [WISPA] Internet Speed test..are they inaccurate with wireless? >> >> We mostly deal with business customer and guarantee bandwidth to >> customers. We validate the bandwidth using IPERF from a Linux server >> off of our BGP edge routers down to the customer and IPERF always >> shows the customer getting the bandwidth they signed up for. We use >> QoS to control bandwidth and make sure to not oversubscribe any one >> linksmall ratios of 3:1. >> >> Of course eventually at some point the customer runs one of those >> stupid bandwidth test on the Internet and the results are woefully >> inaccurate (not in our favor)...but of course customers take the >> results as gospel. AAA! >> >> It's not our internet connections, we have three 100Mbps BGP links and >> none of them run at more then 50% during peak loads. >> >> Has anyone else found those Internet speed test to be woefully >> inaccurate? Or is something else going on that I'm missing? >> >> Thanks, >> Bret >> >> >> >> ___ >> Wireless mailing list >> Wireless@wispa.org >> http://lists.wispa.org/mailman/listinfo/wireless >> >> >> No virus found in this message. >> Checked by AVG - www.avg.com <http://www.avg.com> >> Version: 2012.0.2197 / Virus Database: 2437/5212 - Release Date: 08/20/12 >> > > -- > Scott Reed > Owner > NewWays Networking, LLC > Wireless Networking > Network Design, Installation and Administration > > > > Mikrotik Advanced Certified > > www.nwwnet.net > (765) 855-1060 > (765) 439-4253 > (855) 231-6239 > > > > ___ > Wireless mailing list > Wireless@wispa.org > http://lists.wispa.org/mailman/listinfo/wireless > -- Ing. Paolo Di Francesco Level7 s.r.l. unipersonale Sede operativa: Largo Montalto, 5 - 90144 Palermo C.F. e P.IVA 05940050825 Fax : +39-091-8772072 assistenza: (+39) 091-8776432 web: http://www.level7.it ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Hi Jack you are right about the two points, but also it's not a problem of education but sometimes "honesty" and trustness I will do an example: I turned on some graphs for a customer who was complaining all the time about speed. The issue was that his upload was always full and therefore that was limiting the "user experience" So I explained that and he did not want to believe. So I turned on some graphs to show that the bandwidth was there, and the issue was him not using it in the right way. After some time, he was looking at the graphs all the time and complaining everytime the graph (average on 5mins) were under the "maximum" value (i.e. if the offer was 6Mbit the graph had to be always 6mbit, whatever server he was using). So after all the calls for the "my internet is not going at full speed" I said "ok you know what, I am turning off that feature unless you pay us an extra fee". I was tired of spending my time for no money at the phone with a customer who does not understand "5 min AVERAGE". Moreoever he was saying "it's not full speed, I will not pay the bills" So after turning the graphs off he was complaining that I was hiding some nasty thing because he had no more graphs. So the lessons that I lernt were: 1) trustness: users tend to NOT believe us, even if you give them an accurate, bullet proof explanation. They are not engineers and they simpply say "I did not understand but if I pay for 6Mega I want 6Mega, because with the other WISP I am sure it will go ALWAYS at full speed" (that is the big lie of telecommunication, but who cares...) 2) education: you can spend as much time as you want explaining numbers to them but they will keep only the part they like most of those explanations and use those parts against you (to pay less, to complain, to waste your time, etc.). I am not saying we will stop trying to be polite and explicative with our users, I am just saying that it's an hard path even if you explain with full honesty "why or what" Just my 2 Euro cents > Two points - > > 1. Customer education. I think it's the job of each WISP to educate > their customers that the Internet is a huge network with multiple hops > needed to get anywhere. Yes; I know it seems stupid to us to have to do > this but much of the public is just not aware so we need to help them > understand. > > 2. Local and "Beyond Local" Testing - As part of our customer education, > we need to set up a local (at our NOC) speed test and a "beyond our NOC" > speed test link. That way, we can train the customer to a) check approx. > throughput to our NOC, and b) check approx. throughput beyond our NOC. > > jack > > > > On 8/21/2012 12:22 PM, Bret Clark wrote: >> We mostly deal with business customer and guarantee bandwidth to >> customers. We validate the bandwidth using IPERF from a Linux server >> off of our BGP edge routers down to the customer and IPERF always >> shows the customer getting the bandwidth they signed up for. We use >> QoS to control bandwidth and make sure to not oversubscribe any one >> linksmall ratios of 3:1. >> >> Of course eventually at some point the customer runs one of those >> stupid bandwidth test on the Internet and the results are woefully >> inaccurate (not in our favor)...but of course customers take the >> results as gospel. AAA! >> >> It's not our internet connections, we have three 100Mbps BGP links and >> none of them run at more then 50% during peak loads. >> >> Has anyone else found those Internet speed test to be woefully >> inaccurate? Or is something else going on that I'm missing? >> >> Thanks, >> Bret >> >> >> ___ >> Wireless mailing list >> Wireless@wispa.org >> http://lists.wispa.org/mailman/listinfo/wireless > > -- > Jack Unger - President, Ask-Wi.Com, Inc. > Author (2003) - "Deploying License-Free Wireless Wide-Area Networks" > Serving the WISP Community since 1993 > www.ask-wi.com 760-678-5033jun...@ask-wi.com > > > > > > ___ > Wireless mailing list > Wireless@wispa.org > http://lists.wispa.org/mailman/listinfo/wireless > -- Ing. Paolo Di Francesco Level7 s.r.l. unipersonale Sede operativa: Largo Montalto, 5 - 90144 Palermo C.F. e P.IVA 05940050825 Fax : +39-091-8772072 assistenza: (+39) 091-8776432 web: http://www.level7.it ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Take a look at visualware.com those tests are more accurate than speed test.net Leon Sent from my iPhone On Aug 22, 2012, at 8:26 PM, Blake Covarrubias wrote: > Matt, > > How much bandwidth on average are you seeing to your speed test servers? > > We're considering becoming a speedtest.net host, but are concerned about the > amount of bandwidth generated by users outside our network. > > -- > Blake Covarrubias > > On Aug 22, 2012, at 9:54 AM, Matt Jenkins wrote: > >> We setup a speedtest server. So when customers connect to speedtest.net they >> reach our local server. The test data never leaves our network. >> >> >> On 08/21/2012 12:27 PM, Joey Craig wrote: >>> We only guarantee from the customers radio to our servers and we make sure >>> the customer is aware of that. We use Jperf/Iperf also…works very well. >>> How can you guarantee ANYTHING beyond your network? You are absolutely in >>> the right. >>> >>> Joey Craig >>> Network/RF Engineer >>> Firenet1.Com >>> Phone: (662) 510-0764 >>> Mobile: (662) 404-1118 >>> From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On >>> Behalf Of Bret Clark >>> Sent: Tuesday, August 21, 2012 2:22 PM >>> To: WISPA General List >>> Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? >>> >>> We mostly deal with business customer and guarantee bandwidth to customers. >>> We validate the bandwidth using IPERF from a Linux server off of our BGP >>> edge routers down to the customer and IPERF always shows the customer >>> getting the bandwidth they signed up for. We use QoS to control bandwidth >>> and make sure to not oversubscribe any one linksmall ratios of 3:1. >>> >>> Of course eventually at some point the customer runs one of those stupid >>> bandwidth test on the Internet and the results are woefully inaccurate (not >>> in our favor)...but of course customers take the results as gospel. >>> AAA! >>> >>> It's not our internet connections, we have three 100Mbps BGP links and none >>> of them run at more then 50% during peak loads. >>> >>> Has anyone else found those Internet speed test to be woefully inaccurate? >>> Or is something else going on that I'm missing? >>> >>> Thanks, >>> Bret >>> >>> >>> ___ >>> Wireless mailing list >>> >>> Wireless@wispa.org >>> http://lists.wispa.org/mailman/listinfo/wireless >> >> ___ >> Wireless mailing list >> Wireless@wispa.org >> http://lists.wispa.org/mailman/listinfo/wireless > > ___ > Wireless mailing list > Wireless@wispa.org > http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Matt, How much bandwidth on average are you seeing to your speed test servers? We're considering becoming a speedtest.net host, but are concerned about the amount of bandwidth generated by users outside our network. -- Blake Covarrubias On Aug 22, 2012, at 9:54 AM, Matt Jenkins wrote: > We setup a speedtest server. So when customers connect to speedtest.net they > reach our local server. The test data never leaves our network. > > > On 08/21/2012 12:27 PM, Joey Craig wrote: >> We only guarantee from the customers radio to our servers and we make sure >> the customer is aware of that. We use Jperf/Iperf also…works very well. >> How can you guarantee ANYTHING beyond your network? You are absolutely in >> the right. >> >> Joey Craig >> Network/RF Engineer >> Firenet1.Com >> Phone: (662) 510-0764 >> Mobile: (662) 404-1118 >> From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On >> Behalf Of Bret Clark >> Sent: Tuesday, August 21, 2012 2:22 PM >> To: WISPA General List >> Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? >> >> We mostly deal with business customer and guarantee bandwidth to customers. >> We validate the bandwidth using IPERF from a Linux server off of our BGP >> edge routers down to the customer and IPERF always shows the customer >> getting the bandwidth they signed up for. We use QoS to control bandwidth >> and make sure to not oversubscribe any one linksmall ratios of 3:1. >> >> Of course eventually at some point the customer runs one of those stupid >> bandwidth test on the Internet and the results are woefully inaccurate (not >> in our favor)...but of course customers take the results as gospel. >> AAA! >> >> It's not our internet connections, we have three 100Mbps BGP links and none >> of them run at more then 50% during peak loads. >> >> Has anyone else found those Internet speed test to be woefully inaccurate? >> Or is something else going on that I'm missing? >> >> Thanks, >> Bret >> >> >> ___ >> Wireless mailing list >> >> Wireless@wispa.org >> http://lists.wispa.org/mailman/listinfo/wireless > > ___ > Wireless mailing list > Wireless@wispa.org > http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
http://www.speedtest.net/mini.php On 08/21/2012 12:22 PM, Bret Clark wrote: We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
We setup a speedtest server. So when customers connect to speedtest.net they reach our local server. The test data never leaves our network. On 08/21/2012 12:27 PM, Joey Craig wrote: We only guarantee from the customers radio to our servers and we make sure the customer is aware of that. We use Jperf/Iperf also…works very well. How can you guarantee ANYTHING beyond your network? You are absolutely in the right. Joey Craig Network/RF Engineer Firenet1.Com Phone: (662) 510-0764 Mobile: (662) 404-1118 From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On Behalf Of Bret Clark Sent: Tuesday, August 21, 2012 2:22 PM To: WISPA General List Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
For us this is the biggest one. I will have customer call, speedtest says 384K. Look at the radio and we are pushing 2.04M in on a 2M connection. "Ask your kids to turn off all their devices and try again." "Oh, wow, I am getting the full 2M now. What did you change?" Then the long explanation about shared bandwidth on their connection. Has to be the same for businesses, though it is kind of hard to have all the other kids turn off their computers. On 8/21/2012 3:33 PM, Larry Weidig wrote: Bret: We have found that the tests are "fairly" accurate (well speedtest.net anyhow), *_BUT_* only in controlled environments. Your end user is probably running these tests when their Internet is "slow". Viola look I am right it is slow, xyz speed test says so! What he does not realize is that Johnny intern in the warehouse has set his computer up to grab torrents for him while he picks stock and then just dumps the off to flash before heading out. Ok, that is just one example, but all sorts of other traffic can come into play. We tell our customers that we are not at all interested in speed tests unless we have a support person on the line watching their link for other types of traffic as well during the test. We also host their speedtest.net mini (free) on own of our own servers so that we can eliminate any Internet oddness causing problems with the results. The mini however only seems accurate up to about 25-30 Mbps in our experience at which point the results are only ballpark at best. Ookla claims mini is good until about 50 Mbps, but that has not been our experience. Just what we have experienced with speedtest results. Larry A. Weidig (lwei...@excel.net <mailto:lwei...@excel.net>) Excel.Net, Inc. -- http://www.excel.net/ (920) 452-0455 -- Sheboygan/Plymouth area (888) 489-9995 -- Other areas, toll-free *From:*wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] *On Behalf Of *Bret Clark *Sent:* Tuesday, August 21, 2012 2:22 PM *To:* WISPA General List *Subject:* [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless No virus found in this message. Checked by AVG - www.avg.com <http://www.avg.com> Version: 2012.0.2197 / Virus Database: 2437/5212 - Release Date: 08/20/12 -- Scott Reed Owner NewWays Networking, LLC Wireless Networking Network Design, Installation and Administration Mikrotik Advanced Certified www.nwwnet.net (765) 855-1060 (765) 439-4253 (855) 231-6239 ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
The visualware servers are Accurate and u can deploy them on your network as well. The Dulles,va server is located off the ashburn peering point so that's the one I always use and suggest. Leon Sent from my iPhone On Aug 21, 2012, at 3:22 PM, Bret Clark wrote: > We mostly deal with business customer and guarantee bandwidth to customers. > We validate the bandwidth using IPERF from a Linux server off of our BGP edge > routers down to the customer and IPERF always shows the customer > getting the bandwidth they signed up for. We use QoS to control bandwidth > and make sure to not oversubscribe any one linksmall ratios of 3:1. > > Of course eventually at some point the customer runs one of those stupid > bandwidth test on the Internet and the results are woefully inaccurate (not > in our favor)...but of course customers take the results as gospel. > AAA! > > It's not our internet connections, we have three 100Mbps BGP links and none > of them run at more then 50% during peak loads. > > Has anyone else found those Internet speed test to be woefully inaccurate? Or > is something else going on that I'm missing? > > Thanks, > Bret > ___ > Wireless mailing list > Wireless@wispa.org > http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Two points - 1. Customer education. I think it's the job of each WISP to educate their customers that the Internet is a huge network with multiple hops needed to get anywhere. Yes; I know it seems stupid to us to have to do this but much of the public is just not aware so we need to help them understand. 2. Local and "Beyond Local" Testing - As part of our customer education, we need to set up a local (at our NOC) speed test and a "beyond our NOC" speed test link. That way, we can train the customer to a) check approx. throughput to our NOC, and b) check approx. throughput beyond our NOC. jack On 8/21/2012 12:22 PM, Bret Clark wrote: We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless -- Jack Unger - President, Ask-Wi.Com, Inc. Author (2003) - "Deploying License-Free Wireless Wide-Area Networks" Serving the WISP Community since 1993 www.ask-wi.com 760-678-5033 jun...@ask-wi.com ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
We are lucky as most speedtest.net servers are within 2ms of us. (Probably because we are in two main colos.) Ran one on my desktop at the office yesterday for a prospect and pulled 88 Mbps/92 Mbps from my 100Base-T connected desktop. We are 3 wireless hops from some tests and 5 from others. There is one speedtest.net site that sucks from our network. And it seems to be the one the customers get all the time. Once they get any other server, it's fine. ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Bret: We have found that the tests are "fairly" accurate (well speedtest.net anyhow), BUT only in controlled environments. Your end user is probably running these tests when their Internet is "slow". Viola look I am right it is slow, xyz speed test says so! What he does not realize is that Johnny intern in the warehouse has set his computer up to grab torrents for him while he picks stock and then just dumps the off to flash before heading out. Ok, that is just one example, but all sorts of other traffic can come into play. We tell our customers that we are not at all interested in speed tests unless we have a support person on the line watching their link for other types of traffic as well during the test. We also host their speedtest.net mini (free) on own of our own servers so that we can eliminate any Internet oddness causing problems with the results. The mini however only seems accurate up to about 25-30 Mbps in our experience at which point the results are only ballpark at best. Ookla claims mini is good until about 50 Mbps, but that has not been our experience. Just what we have experienced with speedtest results. Larry A. Weidig (lwei...@excel.net<mailto:lwei...@excel.net>) Excel.Net, Inc. - http://www.excel.net/ (920) 452-0455 - Sheboygan/Plymouth area (888) 489-9995 - Other areas, toll-free From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On Behalf Of Bret Clark Sent: Tuesday, August 21, 2012 2:22 PM To: WISPA General List Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
As my boss would say, "those tests are for entertainment purposes only!" Joey Craig Network/RF Engineer Firenet1.Com Phone: (662) 510-0764 Mobile: (662) 404-1118 -Original Message- From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On Behalf Of Mike Hammett Sent: Tuesday, August 21, 2012 2:25 PM To: WISPA General List Subject: Re: [WISPA] Internet Speed test..are they inaccurate with wireless? You're going to get a bunch of people bashing the tests, but there's generally a reason for a low number. It very well could be nothing on your network, but it's still a problem that needs to be addressed, especially if that is a common testing point in your area. A congested peering point could be to blame. - Mike Hammett Intelligent Computing Solutions http://www.ics-il.com - Original Message - From: "Bret Clark" To: "WISPA General List" Sent: Tuesday, August 21, 2012 2:22:02 PM Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
Well, It really depends. We found them to be extremely inconsistent. We actually host a speedtest.net server just for this reason. We found on very fast connections (Like.. 100Mb/s and above) that they are completely inaccurate. Where is on normal connections (10-30Mb/s) they seem to be pretty close. We take iperf (Or mikrotik speedtest) results over all else. That being said, We generally find when the site based speed tests are extremely low for multiple users there is normally a problem. Even if its hard to see at first. Have this customer test to a few different locations. Then let him download a large zip file from your own network (Like a webserver off your BGP edge routers) and compare. Nick Olsen Network Operations (855) FLSPEED x106 From: "Bret Clark" Sent: Tuesday, August 21, 2012 3:22 PM To: "WISPA General List" Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
You're going to get a bunch of people bashing the tests, but there's generally a reason for a low number. It very well could be nothing on your network, but it's still a problem that needs to be addressed, especially if that is a common testing point in your area. A congested peering point could be to blame. - Mike Hammett Intelligent Computing Solutions http://www.ics-il.com - Original Message - From: "Bret Clark" To: "WISPA General List" Sent: Tuesday, August 21, 2012 2:22:02 PM Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
Re: [WISPA] Internet Speed test..are they inaccurate with wireless?
We only guarantee from the customers radio to our servers and we make sure the customer is aware of that. We use Jperf/Iperf also.works very well. How can you guarantee ANYTHING beyond your network? You are absolutely in the right. Joey Craig Network/RF Engineer <http://www.firenet1.com/> Firenet1.Com Phone: (662) 510-0764 Mobile: (662) 404-1118 _ From: wireless-boun...@wispa.org [mailto:wireless-boun...@wispa.org] On Behalf Of Bret Clark Sent: Tuesday, August 21, 2012 2:22 PM To: WISPA General List Subject: [WISPA] Internet Speed test..are they inaccurate with wireless? We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless
[WISPA] Internet Speed test..are they inaccurate with wireless?
We mostly deal with business customer and guarantee bandwidth to customers. We validate the bandwidth using IPERF from a Linux server off of our BGP edge routers down to the customer and IPERF always shows the customer getting the bandwidth they signed up for. We use QoS to control bandwidth and make sure to not oversubscribe any one linksmall ratios of 3:1. Of course eventually at some point the customer runs one of those stupid bandwidth test on the Internet and the results are woefully inaccurate (not in our favor)...but of course customers take the results as gospel. AAA! It's not our internet connections, we have three 100Mbps BGP links and none of them run at more then 50% during peak loads. Has anyone else found those Internet speed test to be woefully inaccurate? Or is something else going on that I'm missing? Thanks, Bret ___ Wireless mailing list Wireless@wispa.org http://lists.wispa.org/mailman/listinfo/wireless