it sounds like horizontal scaling with redundancy and potentially geographic distriubution on your network would be your big friend here.
On Wed, Nov 11, 2015 at 12:34 AM, Lorell Hathcock <lor...@hathcock.org> wrote: > Good point. There will be no one customer that can get a 10G speedtest from > us. But there will be hundreds that should be able to get a 1G test. > Should any of them try simultaneously, I want to be ready. Plus I don't > know what miscellaneous speedtests from the net to expect, so I want to > affordably overbuild. > > -----Original Message----- > From: valdis.kletni...@vt.edu [mailto:valdis.kletni...@vt.edu] > Sent: Monday, November 9, 2015 11:00 PM > To: Lorell Hathcock <lor...@hathcock.org> > Cc: 'NANOG list' <nanog@nanog.org> > Subject: Re: Updated Ookla Speedtest Server Requirements > > On Mon, 09 Nov 2015 15:27:20 -0600, "Lorell Hathcock" said: > >> I want a server that is capable of handlilng a speedtest up to 10Gbps. > > Do you have (or are planning to have) a clear 10G path to enough systems > that want to use speedtest specifically to make it worthwhile? We have a > lot of gear reachable at high speeds, but the admins of those servers > usually care more about iperf and similar than speedtest..... > >