[tsc] Fwd: URGENT Input Needed for the Technical Strategy Document
David Snelling
David.Snelling at uk.fujitsu.com
Wed Nov 1 10:00:17 CST 2006
Begin forwarded message:
> From: "Franco Travostino" <travos at nortel.com>
> Date: 1 November 2006 14:11:48 GMT
> To: "David Snelling" <David.Snelling at uk.fujitsu.com>, "Cees deLaat"
> <delaat at science.uva.nl>
> Subject: Fwd: Re: URGENT Input Needed for the Technical Strategy
> Document
>
>
> I follow up to the earlier message (see end of this email) with a
> roadmap-worthy update on the set of NM-WG specifications
> anticipated for 2007:
>
> The NM-WG focuses on how to make Grid-relevant network metrics
> available to middleware, network operators etc. NM-WG will not look
> at how measurements should be made. Throughout 2007, the NM-WG will
> produce a set of specifications (standard track documents +
> companion informational/experimental documents) on a V2.0 schema
> for network measurement data and the methods with which to request
> such data. This is a major overhaul of GFD.23 (also known as V1.0).
> The NM-WG has already produced the bulk of the V2.0 work over 2005
> and 2006. The new schema is stable and ready to be finalized into
> the GFD series. Interoperability testing considerations apply (see
> earlier email).
>
> Clearly, the work by the NM-WG has had an impact in the research
> community. Early adopters include EGEE JRA4 and perfSONAR.
> perfSONAR is a joint effort by: ESnet, Fermilab, GÉANT2 JRA1,
> Internet2, and RNP. In commercial settings, the NM-WG is directly
> applicable to any private network (e.g., a Grid within the Boeing
> or Intel worldwide network), though there are no documented cases
> of early adopters. In a provider-provisioned network, however, it
> is unlikely that the provider will agree to expose the kind of
> performance information described in the NM-WG body of work.
>
> As for the overall Infrastructure Area:
>
> The GHPN-RG will move past the use case document (currently in and
> out of Editor's queue) and look squarely into the interoperability
> of Grid Network Service stacks across multiple administrative
> domains, in close cooperation with the Global Lambda Integrated
> Facility organization (www.glif.is). This work will allow a Grid
> infrastructure element to cause the set-up and tear-down of a
> lightpath (an optical circuit) end-to-end across various domain
> boundaries. It will be directly applicable to dark-fiber networks
> (either customer-owned networks or the research & education network
> testbeds).
>
> Throughout 2007. we expect to make progress on the Grid +
> Virtualization front (see earlier email) and the creation of a new
> group concerning with the machine-friendly, Grid-consumable
> description of either virtual or physical topologies.
>
> Cees, anything you want to add?
>
> thanks
> -franco
>
>
>
>> Date: Sun, 29 Oct 2006 10:15:20 -0500
>> To: "David Snelling" <David.Snelling at uk.fujitsu.com>, "Cees
>> deLaat" <delaat at science.uva.nl>
>> From: Franco Travostino <travos at nortel.com>
>> Subject: Re: URGENT Input Needed for the Technical Strategy Document
>> Cc: "Joel Replogle" <replogle at ogf.org>, "Mark Linesch"
>> <mark.linesch at hp.com>
>>
>>
>> Dave,
>> before we provide actual spec updates concerning my Area, I want
>> to brief you on two noteworthy activities that we seeded during
>> calendar year 2006 and could result in material for the roadmap.
>>
>> 1. Grids and Virtualization. As anticipated back in January, we
>> did hold a BOF on this topic at OGF18. The BOF manifesto presented
>> a set of use cases and goals that resonated quite well with the
>> audience, while steering clear of obvious overlaps (e.g., DMTF was
>> represented in the room and helped formulating the charter).
>> Intriguingly, the BOF hit a chord at SAP. Alexander G. of SAP (see
>> his email hereafter) has gotten back to us and is now working with
>> me on drafting an augmented charter. While Alexander's email
>> wanders some into non-OGF turfs (for instance, you can ignore the
>> sentence to the extent that OGF would become a player in the
>> hypervisor space, he later retracted it), Alexander makes an
>> important point. He argues that virtualization is a large enough
>> topic that the OGF governance ought to be involved in isolating
>> the signal from noise and promoting that signal as a
>> distinguishing OGF contribution to the industry.
>>
>> 2. Elements of "Service Grid". The OGF/ITU workshop which we held
>> last week at ITU in Geneva (with Mark) is another key 2006
>> deliverable and a springboard for future activities. The effort of
>> planting the Grid seed over there was worthwhile, despite the fair
>> amount of talking past one another that is due to happen when you
>> put computer-heads and network-heads together in the same room.
>> Specifically, there are elements of Grid infrastructure that
>> seemingly apply well to a SOA-ized next-generation network like
>> the one that ITU is building. These elements include WS-Agreement
>> (with network domain-specific derivations), the security
>> "cartwheel" architecture and supporting specifications, the schema
>> by the Network Monitoring working group, etc. I think of ITU (and
>> others, like the IPsphere consortium) building a "service
>> grid" (this term was introduced by John Seely Brown afaik) and
>> envision that OGF can play a role in this space by way of selected
>> specifications.
>>
>> I will close with a note on interoperability. I want to stick to
>> the optimistic view that the newly formed ETSI TC Grid group will
>> be a meaningful player whose role is to complement OGF. For this,
>> I volunteer that GridFTP and the NM-WG schema are material for
>> plugtest exercises and within the ETSI comfort zone. I would
>> therefore recommend to engage ETSI on these grounds, with GridFTP
>> being the most mature spec in the pipeline.
>>
>> -franco
>>
>>
>>
>>
>>> ---------- Forwarded message ----------
>>> From: Gwen Nichols-White <gnwhite at ogf.org>
>>> Date: Oct 24, 2006 8:21 AM
>>> Subject: DMTF and Virtualization
>>> To: travos at ieee.org, alexander.gebhart at sap.com, "Fogel, Robert" <
>>> robert.fogel at intel.com>, Craig Lee <craig at rushg.aero.org>
>>>
>>> When: Friday, October 27, 2006 10:00 AM-11:00 AM (GMT-06:00)
>>> Central Time
>>> (US & Canada).
>>> Where: Conference Call US-866-875-2241 International
>>> +1-832-445-1174 Code
>>> 2721010
>>>
>>> *~*~*~*~*~*~*~*~*~*
>>>
>>> Pacific-Fri 7:00 AM (Sorry Robert) * Paris-Fri 4:00 PM * Central-
>>> Fri 10:00
>>> AM * Eastern-Fri 9:00 AM
>>>
>>> Conference Call US 1-866-875-2241 International 1-832-445-1174
>>> Code 2721010
>>>
>>> From Alexander:
>>>
>>> Overall Concept / Big Picture
>>>
>>> Virtualization technologies, or to be more specific hypervisor
>>> technologies,
>>> are getting more and more important. Key indicators are
>>>
>>> 1) dramatically increased revenue of VMware ESX
>>> 2) major customers are rapidly deploying virtualization
>>> technologies
>>> 3) estimated VMworld (Nov. 7, 2006) participants: 4000-6000
>>>
>>> Hypervisor technologies are providing key features that are very
>>> important
>>> in a resource management context. Example: VMotion = the ability
>>> to move any
>>> application during runtime from one physical machine to another.
>>> From my
>>> point of view the key capabilities of hypervisors should be
>>> usable by
>>> resource management software, meaning that resource management
>>> software
>>> could integrate hypervisor capabilities. By pursuing this goal,
>>> OGF would
>>> become a player in the hypervisor area.
>>>
>>> Deliverable(s):
>>>
>>> OGF should define what I would call profiles (similar to WSI).
>>> Each profile
>>> consists of a defined set of functionalities and a standardized
>>> interface to
>>> access them. Examples:
>>> Profile A: Functionality = get monitoring information from any
>>> hypervisor
>>> Profile B: Functionality = manipulate hypervisor resources (e.g.
>>> increase
>>> memory, network bandwidth, processing power)
>>> Profile C: Functionality = move VM from one physical machine to
>>> another
>>> OGF could also add best practices / guidance for customers, e.g.
>>> which type
>>> of profile is recommended for which use case.
>>> Another enhancement could be a certification, e.g. OGF certifies
>>> that
>>> resource management system X works perfect with hypervisor Y
>>> using profile A
>>> on HW Z.
>>>
>>> Stakeholders / Key players:
>>> VMware, Xen, IBM, SAP, Intel, Microsoft
>>>
>>> Roadmap:
>>> 1) Identify responsibles. In my opinion it should be an OGF VP,
>>> since this
>>> effort has in my opinion a significant impact on the performance
>>> of OGF as
>>> an organization.
>>> 2) Align with related activities
>>> 3) Get buy-in from the stakeholders / key players
>>> 4) Establish an OGF virtualization working group with the
>>> deliverables
>>> mentioned above
>>> 5) Organize kick-off meeting
>>>
>>
>>
>> At 05:37 PM 10/19/2006, David Snelling wrote:
>>
>>> [Bicycle crash boiler plate: I will include this boiler plate in my
>>> mails for a few weeks to apologize for terseness in the rest of the
>>> mail. I have a broken finger from the crash on Oct 4th, and typing
>>> is hard and possible only in short spurts. Sorry, Dave.]
>>>
>>> Cees and Franco,,
>>>
>>> This has been an outstanding request to all ADs for some time
>>> now, so
>>> I am sending personal reminders to all AD's. I need you to look at
>>> the roadmap table in here and correct the entries for specs in your
>>> area. Also, if you have additional details on each, such as a
>>> list of
>>> those companies/orgs that have committed to implementation by a
>>> certain time, that would help as well. Please aim for the end of the
>>> month, as I know you probably need to contact chairs.
>>>
>>> Just send you responses in email, I'll integrate them into the
>>> document.
>>>
>>> Thanks.
>>> --
>>>
>>> Take care:
>>>
>>> Dr. David Snelling < David . Snelling . UK . Fujitsu . com >
>>> Fujitsu Laboratories of Europe
>>> Hayes Park Central
>>> Hayes End Road
>>> Hayes, Middlesex UB4 8FE
>>>
>>> +44-208-606-4649 (Office)
>>> +44-208-606-4539 (Fax)
>>> +44-7768-807526 (Mobile)
>>>
[Bicycle crash boiler plate: I will include this boiler plate in my
mails for a few weeks to apologize for terseness in the rest of the
mail. I have a broken finger from the crash on Oct 4th, and typing
is hard and possible only in short spurts. Sorry, Dave.]
--
Take care:
Dr. David Snelling < David . Snelling . UK . Fujitsu . com >
Fujitsu Laboratories of Europe
Hayes Park Central
Hayes End Road
Hayes, Middlesex UB4 8FE
+44-208-606-4649 (Office)
+44-208-606-4539 (Fax)
+44-7768-807526 (Mobile)
More information about the tsc
mailing list