Not sure. There were three
connections from various AWS sites pulling bits at
an unruly rate. I've shunted those until congestion
clears.
Probably all just
coincidental.
Denis
On 2021-05-27 7:32 p.m., Jonah Graham wrote:
Thanks Denis for the information. And
well done using the terms and conditions to their
best!
I was wondering if my publishing the
EPP build had an impact as it is a huge release
that presumably needs to rsync to all the mirrors.
If you think it is making things worse, please let
me know and we can discuss mitigation options.
It has nothing to do with peering
or ISPs. We have a 400Mbps bandwidth cap to keep
costs reasonable. All our traffic is priority
over download.e.o since we have mirrors.
By way of coincidence, I've just
recently negotiated an increase to 500Mbps for
minimal cost; this should be coming online
shortly. 25% increase should help.
FWIW, three times each day, at
peak traffic times, we lift the bandwidth cap
altogether and utilize the entire gigabit
uplink for about 20-25 minutes. This goes
unbilled as we are billed at the 95th
percentile, so we make the discarded samples
count :)
Denis
On Thu, May
27, 2021, 4:24 PM Liviu Ionescu, <ilg@xxxxxxxxxx>
wrote:
> On 27 May 2021, at 23:13, Jonah Graham
<jonah@xxxxxxxxxxxxxxxx>
wrote:
>
> nothing to do with being far away
well, actually I'm not that far away, I'm only
176 ms away from Stanford CA: