Whilst we have had some challenges I am very pleased with the outcome and my team handling this, as well as the sales and support people handling customers during all of this work.
Of course, there is always more to do - we have to upgrade and extend links to carriers and transit and peering over the coming months and years as we grow, but I am a lot more confident we can provide the quality of service we have always strived to offer.
I thought some of our more technical customers would like to see this "weather map" overview of a peak usage point even without numbers on it. It is perhaps a rare glimpse at some of our network - the part that provides "broadband" to you all.
It is all about us not being the bottleneck and not dropping packets.
Thanks, fascinating stuff! From the earlier post, each BT link was assigned to a specific LNS, while TT links floated between them (so TT-BT bonding could work), so there's actually a switch or two joining those links together too - the same core switches, with each FB's two ports bonded across the pair of switches for redundancy/resilience?
ReplyDeleteNice to know that now scaling is "just" a matter of scaling out with more Firebricks, more peering/transit links etc. At least as long as you don't get more hoodies ram-raiding the office again!*
It's a handy diagram, perhaps something like it could enhance linkreport.cgi when someone has time? The Be links have gone from the diagram, probably time they disappeared from the text too...
(* The first "real" web server I ran was stolen in a datacentre breakin back in 1999, which was a shame. I wonder how much "hot" Sun Enterprise kit changed hand for in the pub in those days?!)
I'd be interested in you writing a bit about peering, if you get the time...
ReplyDelete