Re: [Beacon] spec feedback + few suggestions

Would that mean that the total allowed queued would essentially depend on
transmission rate? If you can send 64Kbps, then you can send 6.4 10K
beacons per second? Or would there still be some kind of limit on how many
beacons can be sent per unit of time?


On Tue, Nov 19, 2013 at 10:07 PM, Jonas Sicking <jonas@sicking.cc> wrote:

> On Tue, Nov 19, 2013 at 5:08 PM, Ilya Grigorik <igrigorik@google.com>
> wrote:
> >
> > On Wed, Nov 6, 2013 at 4:11 PM, Jatinder Mann <jmann@microsoft.com>
> wrote:
> >>
> >> I’ve also opened Action 111: Determine appropriate size limit for beacon
> >> data based on current usage patterns [2] to help us determine what the
> right
> >> limit should be for beacon data transmissions and whether we should set
> a
> >> limit on number of beacons per page or rate at which beacons are
> >> transmitted. Ilya, you had mentioned that you were going to look at
> Google
> >> Analytics data to see what the common analytics package sizes are? That
> data
> >> will go a long way in helping us determine an appropriate limit.
> >
> > After some back and forth with the GA team... "10KB per beacon limit is
> > sufficient for current GA needs". Unfortunately, the GA team has asked
> me to
> > not disclose the exact distribution, but based on the data I've seen,
> 10KB
> > is in the right ballpark: it covers current use cases and provides some
> > margin for growth. That said, values below 10KB will begin to pose
> problems
> > pretty quickly.. So, it may be worth considering 10KB as a good floor
> value.
>
> Having talked with the mozilla network people, I would recommend a
> 10Kb per beacon limit, and a 64Kb total queued beacon data per origin
> limit. I.e. a website would be allowed to queue up as many beacons as
> it wants, but the accumulated size of all beacons must be less than
> 64Kb.
>
> / Jonas
>