[clouds] IETF-77 Clouds bar BoF mtg minutes for your review and comments, if any

[Bhumip Khasnabish <[e-mail]>]

 Dear All,

Below please find the IETF-77 Clouds bar BoF mtg minutes.

Please let me know if you have any comments.

Thanks a lot for quick response.

Bhumip (Mobile:[phone])


=================================START======================
A Clouds bar BoF was held during the IETF-77 on Thursday, March 25, 2010 at
11:30am.
The meeting was held at Pacific B Room at the Hilton Anaheim.
The bar BoF convener was Bhumip Khasnabish ([e-mail]).

Bhumip started the informal discussion with a few slides with purpose to
then spend the meeting having an open discussion.  There were 81 attendees
at the clouds bar BoF.

Bhumip gave a summary of what a Cloud-based system is.  He then provided
slides on service over Cloud.

Bhumip stated that there is no Standard Reference Framework for Cloud
Services.  He concluded with a suggestion to conduct a survey of Cloud-based
systems and services and prepare a report.  He noted the need to synergize
activities with what are being done in other groups such as DECADE,
IRTF/vnrg, NFSv4 and other groups.    Finally, Bhumip stated that a mailing
list for these informal discussions was created called [e-mail].
After Bhumip short slides an open discussion was started so that attendees
can share thoughts on such topic areas.

Open Discussion:
Hannes Tschofenig (Nokia Siemens):   Asked those in room who of you
participate in other SDOs related to clouds.

Scott Brim (Cisco):   DMTF Open Cloud Standards Incubator.  Scott stated
that DMTF is an Incubator Launched to Address Management Interoperability
for Cloud Systems.  Scott stated that the incubation process is designed to
foster and expedite open, collaborative, exploratory technical work that
complements the DMTF mission to lead the development, adoption and promotion
of interoperable management initiatives and standards.  Scott gave example
of some companies that participate in DMTF:   Cisco, EMC, HP, Microsoft,
VMware, Novell, etc.

It was also mentioned that the Clouds Security Alliance (CSA) also exist.
It was stated that the mission of the CSA is to promote the use of best
practices for providing security assurance within Cloud Computing, and
provide education on the uses of Cloud Computing to help secure all other
forms of computing.  It was stated that there are many companies who
participate in CSA including Cisco, Google, Verizon, HP, VMware, Microsoft,
AT&T, etc.
Paul  Hoffman (VPN Consortium):    Noted XML digital signature and
encryption.   IETF can bring some “don’t
do that” but that doesn’t make a good charter.
Cullen Jennings (Cisco):  We can mandate that they use IPv6.
Monique Morrow (Cisco):  Asked the question if you have dealt with any ‘gap’
analysis?
Bhumip: IETF can define and use adaptive protocol for use between virtual
and physical resources, e.g., protocol between virtual network and
real-network.
Spencer:  When you report out, put ‘gap’ analysis in it.
Spencer:  if this work comes into the IETF, differences of building system
and figure out what charter is for work from a ‘gap’ analysis.
Hannes:  Brief chat from Cullen.  Do we have stake holders here?  For
example, VMware?
Paul Hoffman:   VMware, Cytrix, etc , are active in many of these things but
the major players haven’t come to ask for openness.  This is of concern.
Paul:  We may have a problem that doesn’t need to be solved from a customer
point of view.
Dave Crocker (Brandenburg InternetWorking):  There may be some problems that
are useful that others aren’t doing.   Make sure that there are consumers
for this work.  Be clear that there is some group in market that wants this
work.  For example, open hybrid support mechanism.  Cloud-bursting… don’t
know if there is a market for it but there is interest for it.
Spencer:  We get more comfortable with it when potential real consumers show
up.
Cullen:   Unified information group:  move into clouds type environment.
One topic found – if have elastic band to add servers.  DYNDNS.. how to add
more servers .  DNS was a topic we found lacking.
Paul:  We do have APIs out there.  Look at API… Sun case – what are the
linkages.  Hardest part, if … immediate look at the same system it is on,
should we do DHCP – single RFC… VMware is not open API, command line
interface with VMware, that might do good place to do the “gap” analysis.
Carl Moberg:  Look at gaps in Internet architecture.  An example is to
measure proximity and I am not talking about geographic proximity.  Because
sometimes things in clouds need low latency, low jitter, etc.  Sometimes you
don’t care about low latency.  So you need a proximity measure tool. When
wondering around in this cloud space, who is my best peer to bind to.   We
don’t have good tools for that right now.   Things in cloud split and
coalesce and move – and in this case we need naming system – to express
continuous relationship pieces of an entity for example when it splits.
Need dynamic naming system.
David Bryan (Cogent Force):   Noted the topological order of peers – some
coordination with the groups that deal with peer-to-peer may be useful.  It
spans.
Spencer:  Will ALTO do what you need?
Carl:  Not familiar with that work.  Metric I was looking for is it being
lightweight and get it fast.
Yunfei (China Mobile):  Can you explain private cloud, public cloud.
Bhumip:   Things are still being defined.  Not defined in standards. Many
people are talking about these private, public, and hybrid clouds.
Yunfei:   Two types of resources: physical and virtual; many of clouds
deployments use virtual machine to allocate physical resources.
Bhumip:  A survey is being planned; virtual resources have to be mapped to
physical resources, and the interfaces that expose/hide the physical
resources need to be defined and standardized.
Yunfei:  Given so many proprietary methods on virtual machines, this may be
hard.
Dave Crocker:   Now go back to problem statement slide that Bhumip showed.
Everything is nice to list.  Those are reasonable things to list.  In terms
of actual work, it is doesn’t seem to go very well.
I was trying to think how to get work started.  Anything that attacks
problem in larger will likely fail.  Stop using the word cloud.  It has no
meaning because it has too many meanings.
Clouds today use IETF technologies.  They have problems in the use of that
technology.  IP address flux and DNS flux are examples.  Finding some of
these which are narrow problems and looking for ways to enhance them or to
provide alternatives might be a well to get concrete work done and have
significant utility to the cloud community.    Doing this we can avoid
pursuing gaseous topics.
Paul:  Follow-up on Dave Crocker.  What it sounds like to me is that cloud
problems that are already here don’t need working groups as they may already
have working groups.   As a full virtual machine is moving around possible
interface or universal interface and Mobile IP may not be the answer to that
problem.  It may be that in fact that we may get started on some of these
things by saying to these working groups “hey we are hearing that there are
problems with this, can you start working on this with the new assumption
that these objects can move around faster than what we see in the stable
world or the traditional cellular world….”  That may take a little pressure
off of us in forming a working group to cover all these things.
Erik:  If you fan things out to working group, then you need to understand
the gap analysis first.
HuiDeng: Have you looked at what OCCI is defining?
Pengjin: About the problem statement, what want to do for this work, what
want to define for the cloud computing?
Hannes:  Case is smartgrid.  Lots of excitement.  Plugged together in a
reasonable way.  What is missing is a use case description.   Virtual
machines that move around.  Other cases that require….  Have you looked at….
Bhumip:  One example is ‘Infrastructure as a service.”
Monique:  look at other SDOs have definitions and operators have already
started implementing them.  There could be a ‘gap’ analysis.
Paul:  Once we start doing ‘gap’ analysis, use our IETF draft system to do
it.
Paul:  Haven’t seen a ‘gap’ analysis on this topic.
Bhumip:  First we’ll do the ‘gap’ analysis and then a survey on cloud
services. We can look into what other SDOs we can refer to. Let us discuss
these further in the next meeting during IETF-78.
====================END of IETF77 Clouds bar BoF meeting
minutes======================