* Re: [Bloat] what is ready for prime time?
2011-11-23 0:33 [Bloat] what is ready for prime time? david
@ 2011-11-24 11:41 ` Dave Taht
0 siblings, 0 replies; 2+ messages in thread
From: Dave Taht @ 2011-11-24 11:41 UTC (permalink / raw)
To: david; +Cc: bloat
Dear David:
I haven't been ignoring your email, it's just caused me to reflect upon
what we have and have not accomplished this year with fixing
bufferbloat on servers, desktops, and cpe.
So I've been slowly writing up overall 'lessons learned'
writup as sort of a close for a long year in fighting bufferbloat.
Mentally, the projects' sort of to the tune of "alice's restaurant", and
this is a really good illustrated version of that linked to below.
http://www.youtube.com/watch?v=JmyXTOHC3w8
I would suggest you stop reading further (because I'm about
to wander off topic, and I swear this is really wonderful, and just
dig that for a while)
Anyway...
Part 1 of me trying to figure out what we were doing right/wrong
got posted up here...
http://the-edge.blogspot.com/2011/10/10-suggestions-for-startups.html
I keep applying startup-like approaches to what is an R&D problem and
not intended to be a startup (in fact a charitable org, if anything)
... and keep shooting myself in the foot. Some amount of funding from
somewhere would help tackle the really hard problems like testing at
scale, and developing better test tools, and so on and so on.
Startup-mode just happens to be my primary background for solving big problems.
I wish I could figure out a good way for bufferbloat.net
to be less of a 'success disaster'.
Jim and I'd felt, at the beginning, that since academia hadn't worked, and
big industry hadn't worked, and the end users didn't know what was
going wrong..,
that we'd try open source, a full court 'press' effort, crowdsourcing,
and anything
else we could come up with to try and get latency over the entire internet back
under control before it maybe spiraled out of control, which still
keeps jg and I awake at night...
and created this as a mission statement:
"The Bufferbloat projects are providing a webspace for addressing
chaotic and laggy network performance by:
Gathering together experts to tackle networking queue management
and system problem(s), particularly those that effect wireless
networks, home gateways, and edge routers
Spreading the word to correct basic assumptions regarding goodput
and good buffering on the laptop, home gateway, core routers and
servers.
Producing tools to demonstrate and diagnose the problem
Experiments in advanced congestion management
Producing patches to popular operating systems at the device
driver, queuing, and TCP/ip layers to fix the problems."
And ya know, we're meeting those goals to a large extent.
Of late tom Herbert's work on BQL looks like a win...
we have got a bunch of wonderful volunteers, and donated resources
from all over the world, notably from isc.org and lincs.fr, multiple
folk at various levels in multiple big companies grok the problem,
and we have all sorts of folks, academics and engineers, cats and dogs all
living together, in harmony...
(cue the song alice's restaurant (it's thanksgiving after all!))
I'm really grateful for all that.
Thanks everybody! I'm extremely open as to suggestions about what we
can do next year.
BTW: This is a really great version of alice's restaurant, seriously...
http://www.youtube.com/watch?v=JmyXTOHC3w8
anyway, back to your question...
In regards to the cerowrt sub-project... I have high hopes
that rc8 will be stable by january and actually show an improvement
in nearly every respect over the factory firmware.
But:
On my bad days I look at the closed bugs to cheer me up.
On my good days I look at the open bugs to bring me down.
I have more bad days than good.
It may not stablize, either. Although I can say that we've
generally 'won' on stability in the first place...
So I have a detailed list of lessons learned this year that
may apply to your deployment at SCALE in various ways,
on top of any os - openwrt, cerowrt, cisco, linux, etc...
but it's currently folded into the above much longer and
fragmented list...
I would be curious however to obtain a network map
of what you tried last year, and how you already plan to
solve some of it this year.
(I'm sort of being socratic in that I'm very interested in how
well we've communicated anything in the last year, and
learning how well that truly has made it 'out there' is
rather high on my list of things I don't know.
For example, what sort of hardware do you plan on using
for the 45Mbps link?)
Happy thanksgiving all! I'm logging out for a few days. I'll get
back to answering your underlying question on monday.
"You can get anything you want at alice's restaurant'.
On Wed, Nov 23, 2011 at 1:33 AM, <david@lang.hm> wrote:
> I'm running the wireless network for SCALE in late January (~1800 people
> with a 45Mb uplink), we will be using ~30 netgear WNDR3700v2 access points
> running openwrt (or a derivitive). What of the stuff that is in CeroWRT is
> considered reliable enough for production use like this? (I know that a lot
> of stuff there is not)
>
> Also, what would you folks like to have me put on the APs or network to
> gather info under these harsh conditions?
>
> last year we moved to a larger hotel and I ran fairly stock DD-WRT on the
> same access points I had used the year before and things did not work very
> well, but I think I ended up without having dense enough coverage so this
> year I'm moving to openwrt (to give me more control) and doubling the number
> of access points.
>
> David Lang
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
>
--
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
FR Tel: 0638645374
http://www.bufferbloat.net
^ permalink raw reply [flat|nested] 2+ messages in thread