On Sun, Mar 11, 2012 at 4:50 AM, Jeremy Visser <jeremy@visser.name> wrote:
On 11/03/12 10:43, Dave Taht wrote:
A ton of people offered help at the time, but they all offered
virtual boxes in places that had no native ipv6. It's kind of my hope
that native ipv6 support in US data centers has improved since then?

Does poo-pooing VPSes with non-native IPv6 also exclude VPSes with native IPv6?

If not, Linode is the option for you. They have DCs (Atlanta, GA or Newark, NJ) on the East Coast with native IPv6.

It was kind of my hope to get this off the ground in april.

I have two conflicting desires:

1) be able to do create a conference server as useful infrastructure for supporting this project, with things like irc support and other oddities. My requirements for this was native ipv6, a conference server, some sort of gui, no pots gw was required.

As the east coast would have the least latency overall, locating it there was highly desirable....

but then, there's:

2) be able to observe and fix problems at the ip layer with the new  bql/aqm stuff against voip and tcp traffic in the real world....

I'm a big believer eating our own dogfood; I intend to update all of our main servers to run linux 3.3 with sfqred enabled (after another month's worth of testing, I'm crazy, but not that crazy),

... but I was not big on the idea of adding a voip server to  huchra, or taking another piece of test gear (io.lab.bufferbloat.net) out of experimental and into a more production status, and in both cases, those boxes are located on the west coast.

I HAVE been able to coax asterisk and freeswitch to work on openwrt in the past but somehow I doubt it's suitable for a conference server... it would be interesting to try as a conventional switch, but... it's on a conference server that problems become most readily apparent.

 So at the moment, I'm thinking the best idea is to put up a test server on the west coast to meet desire 2, and work towards a way to apply desires 2 and 1 on the east coast, eventually.

In both cases I'm still kind of allergic to vms because that just eliminates a whole layer of control over the stack that we'd have if we  pursued and abused the latest kernel. I would certainly like to observe vm behavior, but not for 3-6 months longer, after the underlying OS can also be updated to 3.3....


_______________________________________________
Bloat mailing list
Bloat@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/bloat



--
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net