We also setup a Speedtest.net (Ookla) public speed test server at Notre Dame. Our main motivation was to manage perception. We are on a state run optical network. Our speedtest.net traffic went from campus on the north end of the state, to central Indiana, to Chicago, then back to South Bend. The closest geographical Speedtest.net public server was already in our town, but due to our ISP setup there was a lot of excessive travel. The closest geographical test server did not appear to be on a fast enough link either. We unsuccessfully tried to get Speedtest.net to point our public IP space to Indianapolis speedtest.net servers to get a more accurate test results. I see Comcast and AT&T are able to point speedtest.net to the closest logical test server instead of closest geographical test server. We ended up installing the Speedtest.net free public server. Without the paid subscription we do not get access to detailed information on test results. Less detail was fine for us because we just needed to handle the perception issue caused by speed tests going around the state(s) to a less optimal test server. We also setup a lightweight server http://speedtest.nd.edu but found a lot of students prefer (trust) third party test results from sites they use at home. Even some faculty will use speedtest.net as a quick check prior to setting up iperf or perfsonar. It is quick and easy…. If the results look okay they move on to solving the world’s problems instead of building infrastructure to test our infrastructure. Which circles back to getting users to trust your infrastructure simply because of a test result that used to be out of our scope.
Side note, HDD speed affects Ookla speedtest.net server performance. We ended up putting an NVMe drive into the old repurposed server to better serve multi gigabit connections. *Mike Atkins * Network Engineer Office of Information Technology University of Notre Dame *From:* The EDUCAUSE Wireless Issues Constituent Group Listserv [mailto: [email protected]] *On Behalf Of *Amel Caldwell *Sent:* Monday, February 26, 2018 10:59 AM *To:* [email protected] *Subject:* Re: [WIRELESS-LAN] Bandwidth/Throughput/Latency Tester We also have an instance of the Ookla speedtest at the University of Washington. One thing I notices is for clients on private IP space, the speedtest shows a NATed IP, even though the server is on campus. This is because not everything is local. Anyway, having someone send me a screenshot or tell me their IP address is the NATed address is not that helpful. I believe we are considering an alternative when our year is up. Amel Caldwell University of Washington UW-IT Wi-Fi Network Engineer Wi-Fi Service Manager [email protected] 206-543-2915 *From: *The EDUCAUSE Wireless Issues Constituent Group Listserv < [email protected]> on behalf of "Osborne, Bruce W (Network Operations)" <[email protected]> *Reply-To: *The EDUCAUSE Wireless Issues Constituent Group Listserv < [email protected]> *Date: *Monday, February 26, 2018 at 4:56 AM *To: *"[email protected]" < [email protected]> *Subject: *Re: [WIRELESS-LAN] Bandwidth/Throughput/Latency Tester That is what we use. http://speedtest.liberty.edu *Bruce Osborne* *Senior Network Engineer* *Network Operations - Wireless* *(434) 592-4229* *LIBERTY UNIVERSITY* *Training Champions for Christ since 1971* *From:* Adam Forsyth [mailto:[email protected] <[email protected]>] *Sent:* Friday, February 23, 2018 9:53 AM *Subject:* Re: Bandwidth/Throughput/Latency Tester Isn't this: https://www.ookla.com/speedtest-custom what you asked Ookla about and were told that it doesn't exist? I ran a version of that on a local server a few years ago.I got the premium subscription for a year but ultimately decided I hadn't figured out how to get any advantage from its ability to save test results into a database. I have since moved to using https://github.com/adolfintel/speedtest (which Clemson also mentioned) because I wanted a speedtest that was HTML5 and didn't use flash, and at the time Ookla's speedtest custom required flash. It looks like maybe its also all HTML5 now so maybe I'll take a look at that again. On Tue, Feb 20, 2018 at 11:56 AM, Fishel Erps < [email protected]> wrote: Hello everyone. I’m curious to find out what other universities are doing to test throughput, internally, to proof their networks. I’m looking for something that functions like Ookla’s Speedtest.net (browser-based, no required clients) , but that runs internally (I have already contacted them directly, and been told that they only provide products that are alive on the public net). As we all know, % of utilization and available throughput are not one-in-the-same, and I need a way to address and diagnose legitimate performance complaints, live. __________________________________ __________________________________ Fishel Erps, Sr. Network & Infrastructure Engineer School of Visual Arts 136 W 21st St., 8th Floor New York, NY, 10011 LL: 212-592-2416 F: 646-845-6150 E: [email protected] _______________________________ Please excuse any typographical errors as this e-mail has been sent from my mobile device _______________________________ ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/discuss. -- *Adam Forsyth* Director of Network and Systems Luther College Information Technology Services 700 College Drive Decorah, IA 52101 563-387-1402 ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/discuss. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/discuss. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/discuss. ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/discuss.
