[Bloat] a request I made to ISOC and IETF TSV AD

[Michael Richardson]

this is a FYI.

ISOC says they are full for 2013.  If someone can think of another
entity to coordinate, and take some of the *outreach* load off
of JG, DT, and friends, name it..

On May 6, 2013, at 3:49 PM, Michael Richardson <mcr+ietf at sandelman.ca> wrote:
> Welcome to the IESG, Spencer.
> 
> I am putting my <operator> hat on as <mcr at novavision.ca>.
> Novavision is a boutique business ISP in Montreal, Quebec has fiber in a
> number of industrial parks.  I'm the network architect and director of R&D.
> We have serious bufferbloat issues that prevent us from deploying the
> kind of service we want.  In many cases I control both ends of the
> layer-3, and I could deploy whatever I want.  If only I could train the
> sales engineers of my vendors... 

> Awhile ago I suggested to Dan York that ISOC should consider adding
> bufferbloat to it's Deploy360-ish efforts.  A key thing for me around
> bufferbloat is:
>  a) convincing companies that it's real.  A video hosted by
>     Vint or Bob Kahn, aimed at semi-technical CTOs would help.
> 
>  b) having convinced them that it's real, I need to find out what
>     they are doing about it, and what work arounds they might have.
> 
>     I suggest a well curated wiki for (b), with encouragement for
>     vendors to link directly to their "knowledge bases", etc.
> 
>  c) some point of contact for bufferbloat issues... 
>     This would have to start with some kind of IETF led attempt to 
>     actually find out who at various companies might be in charge
>     of figuring who is in charge of figuring out what the contact would
>     be.
> 
> Some small vendor specific background.  Novavision is a Juniper partner.
> I explained that I couldn't buy a product until I had some clear
> statements about bufferbloat plans from the sales engineer.   I tried to get
> the SE some contacts... I used various contacts I had @juniper.net, but
> they came up blank.   They didn't know how to address this question
> either.   I think that this is a industry wide problem.
>