On Feb 8, 2008, at 12:30 PM, Iljitsch van Beijnum wrote:
I suggest we start thinking about this now rather than at the point
where the IETF can't pay its bills anymore. Where do we draw the
line on meeting fee increases? Is there any way to save costs? What
was the cost structure 10 or 15 years ago when meeting attendance
was much smaller?
Thanks. Yes, the IAOC is thinking about that too. We spent Monday and
Tuesday this week in Helsinki on exactly that question. The IETF
didn't cover our travel :-)
To be honest, a lot of the data regarding meetings 15 years ago isn't
available. We don't know what the costs were. "Better financial
insight" was one of the reasons we went through the process of moving
the secretariat away from CNRI in the first place. I'll also bicker
with your numbers a bit; in March 1998 we were comparable in size to
what we do now, and we were spending the last of our NSF subsidy
funding. In 1993 we had our first meeting in Amsterdam, ~700 people
in the RAI conference center, funded in part by US research funding.
I'm not sure we're apples and apples in this.
Before 1992, the IETF largely met in academic donated meeting
facilities. When the meeting went above ~500 people, that became
untenable, and we have used hotels and conference centers since. One
option is to look again at donated facilities. We have asked hotels
to provide a vendor selling sandwiches for lunch; we could have a
vendor sell coffee and cookies. One thing I'm hoping AMS can do
better for us than CNRI/Foretec/Neustar did is meeting cost
negotiations - I suspect there is still some blood in that turnip.
One thing the IAOC is looking at at this instant is our phone bill.
The IETF's phone budget for 2008 is
Now, our sense is that we can improve meeting quality for the above
and reduce the cost by using a VoIP service. One that comes quickly
to mind is Skype; that has some issues, both with confidentiality and
Net Neutrality. It is also currently limited to conferences of 5
participants, 10 participants if everyone has a dual-core processor.
For an IESG meeting, the implied peak bandwidth is quite high -
19*70=1330 KBPS - because the data would be sent to 19 other
participants unicast (the IESG is 20 people). My service is 384 KBPS
upload, so I personally would have to pay a lot more money if I were
to volunteer for the role. I don't think we're going to use Skype. So
we're looking for something that either uses a central server and
mixes sound, or runs IP multicast over the backbone, which is (ahem)
not a ubiquitous service. The ISOC Board uses a webex-like package
called Marratech (http://en.wikipedia.org/wiki/Marratech), which has
since been acquired from Marratech by Google. ISOC purchased the
necessary server licenses and can expand them if needed; we think
this part of the budget can be dramatically reduced by asking
participants to download the free software and use it. The bodies
we'd be asking haven't yet had that discussion, but its something the
IAOC is looking at.
We're looking pretty hard at the RFC Editor contract, which has a
large overhead fee built into it. Stay tuned in that regard. We have
some ideas and will be doing an RFI or RFP later this year, but they
aren't sufficiently baked just yet to pass aromas around.
I won't go through the budget line by line, but you get the idea. In
a $4.9M budget, we are looking at a few line items in 6 digits and a
number more in five digits, and asking in each case how to change N
digits to N-1.
Ietf mailing list