DMTF to Develop Standards for Managing a Cloud Computing Environment

[Mark A. Carlson]

Open Cloud Standards Incubator Created to Address Management Interoperability for
Cloud Systems
 
PORTLAND, Ore. – April 27, 2009 – The Distributed Management Task Force (DMTF), the
organization bringing the IT industry together to collaborate on systems management standards
development, validation, promotion and adoption, today announced that it has formed a group
dedicated to addressing the need for open management standards for cloud computing. The
“Open Cloud Standards Incubator” will work to develop a set of informational specifications for
cloud resource management.
As virtualization technology continues to be more rapidly adopted, it is emerging as a common
enabling foundation for delivering software solutions into IT environments along with the
potential to lower IT costs and improve operational efficiencies.  While deploying virtualization
technologies it is also critical to have comprehensive management capabilities associated with
the implementation.  Along with the adoption of virtualization, more and more enterprise IT
customers are looking at the cloud computing paradigm to better deliver services to their
customers. 
No specific standards currently exist for enabling interoperability between private clouds within
enterprises and hosted or public cloud providers. DMTF’s Open Cloud Standards Incubator will
focus on addressing these issues by developing cloud resource management protocols,
packaging formats and security mechanisms to facilitate interoperability.

For more info, see: http://dmtf.org/cloud

-- mark

[Reuven Cohen]

Mark, thanks for sharing the news.

Also interesting to note that the "Open Cloud" Standards Incubator is
being led by AMD, Cisco, Citrix, EMC, HP, IBM, Intel, Microsoft,
Novell, RedHat, Savvis, Sun Microsystems, and VMware. What you will
noticed is there are no smaller startups or "cloud specific" companies
such as Google or Amazon included. At first glance this looks like a
list of the old boys of tech. I can't help but wonder if this will
help or hinder the adoption. An open development process is critical
to the success of any standards that may get created by this group.
The last thing we want is to be forced to adopt a set of technical
standards because our partners are telling us we need to. I'd rather
see broad adoption from the smaller and bigger players alike as the
driving factor when looking to implement a new set of technology
standards.

How can we as a community help with this initiative? Are all
contributors required to join the DMTF? Also, I didn't see in any of
the press release info what the time frame is for the development of
these standards?

Ruv

[Mark A. Carlson]

Reuven,

Think of this as the Virtualization vendors getting together to standardize on how they
address the cloud with their products. If you look at the list, you see pretty much every
virtualization vendor participating.

If this group forms a Cloud trade association, I can see us establishing an alliance with
the DMTF to coordinate the messaging and driving adoption of the DMTF standards.
One of the hats I wear is the DMTF VP of Alliances, and I would be happy to help
facilitate such an alliance. The DMTF already has alliances with SNIA and OGF who
are also working in the cloud space.

-- mark

--

Mark A. Carlson Sr. Architect Systems Group Phone x69559 / 303-223-6139 Email Mark.C...@Sun.COM

[Sam Johnston]

Reuven,

The most recent draft of the Open Cloud Principles was loosened up a little to cater for specs like Amazon's EC2 and Google GData, albeit not entirely without contention so we may see that pulled back a little in the final version. The best standards tend to be developed by way of an IETF/OGF style open process in which any interested party can get involved while the worst standards tend to stem from a single vendor. Multi-vendor efforts like this sit somewhere in between the two extremes - the going rate for a vote at DMTF is currently $12,000 though there are various levels of participation down to a $200 "sponsored" individual - early access by way of a "monitoring" membership will cost you $2,500 a year.

Technical aspects are only half of the battle though - marketing and politics are equally important and at the end of the day implementation and adoption are all that matters... despite having IP (copyright, patent and trademark) concerns, the Amazon EC2 API is for better or worse the de facto standard for cloud computing today.

The smart money's still on the Open Cloud Computing Interface (OCCI) as it will deliver an implementable draft based on a "consensus" of existing APIs (including Amazon, GoGrid, ElasticHosts, etc.) within a month and should have a number of interoperable implementations available some time before the DMTF specs are public. It is also somewhat more focused on public cloud infrastructure services than virtualisation-based "private clouds" and has strong representation from both academia and end users rather than being a vendor-only effort. You're welcome to get involved over at http://www.occi-wg.org/ and the next conf call is Wednesday.

Sam

[Alex Esterkin]

Reuven,

Great point!  I agree with you 100%.

By the way, note that Google and Amazon have not yet joined the "Cloud Manifesto".      

Frankly, I am surprised at how exclusive this standard development effort appears to be.  Just look at the membership levels description page: http://www.dmtf.org/join/levels/   There is no place for emerging vendors or experts.   You have to pay $2500 annually just to be able to read early drafts and $6000 to submit comments.   The price tag and "big boys of tech" dominance will guarantee that by the time a new standard is released, the "big boys" will have all the necessary related products and tools already developed, tested, and ready for shipping.

I would like to see something resembling the Java Community Process, something very inclusive, fostering active collaboration of all stakeholders - big companies, small companies, startups, commercial, government, non-profit, and academic users, integrators, developers, experts, and academics. 

So much for "Openness" in the "Open Cloud". 

Regards,
Alex Esterkin

[David Bernstein (daberns)]

Hi All

 

As one of the participants in the DMTF effort I’d like to comment. First of all, the DMTF is a class-A, professional standards organization with no hidden agendas and no desire to shut anyone out. DMTF was formed as a “classic” standards organization and the members have worked hard to produce technologies which generally have been good contributions to improving the state of the art of open systems computing over the years. I truly believe that the $200 individual membership level is perfectly reasonable if you want to get involved.. I am not going to argue whether the membership and voting rights levels and system is perfect or not but I can tell you that  there’s no intention on behalf of the DMTF to shut anyone out, and the people working on these projects are just guys trying to develop interoperable technology which works across multiple implementations.

 

I’ve posted many times here, that in order for people and companies small to large to work together on technology, some form of formal organization with a legal structure is required. And it does take money to run one of these. Organizations from IETF to JCP to DMTF all employ these in some form or another. You may argue that the IETF was constructed better in some ways or the JCP is in other ways but as a participant over the years in all of these I can tell you, none of them is perfect and they all have complications. If anyone is going to stand up here and say that the JCP does not have it’s own share of issues then I’d be very very surprised, you probably didn’t actually work much in that venue. Likewise the IETF. Additionally, they have areas of concentration. JCP is simply not looking at hypervisor virtualization at the file formats sort of level (not APIs). Nor is IETF. Or OGF. DMTF is and as Mark from Sun pointed out, there is a very nice critical mass of  “Virtualization vendors getting together to standardize on how they
address the cloud with their products”.

 

So folks like myself, and Sun and IBM and VMware and Citrix, and RedHat and ….wanting to put actual engineers and time and money into trying to further the interoperability and standards around hypervisors/virtual machine technology at the file formats level, we are working with good intentions in DMTF, I can assure you it is not for a bunch of sneaky reasons, it is because it is the right thing to do.

 

We will do the same in SNIA around storage efforts, and if OGF OCCI has some interesting API progress and gains some other significant members, well OGF is also a terrific class-A SDO, we’ll put some contribution there too.

 

Please support the DMTF effort. If you want to participate, and can’t be a full member, eg you need a sponsor to apply for the $200 individual membership, I am sure any of the member companies including mine, will be happy to oblige.

 

David Bernstein

Cisco.

[Reuven Cohen]

I had a great chat last week with Winston Bumpus at the DMTF, he gave
me a great overview of the DMTF's current plans around cloud
computing. For the most part the DMTF has a great roadmap. I also
mentioned the lack of community involvement and he said that they're
open to the idea.

First let me say the DMTF has the legal structure and industry
participation required for broad adoption of any cloud standards, but
lacks any real community involvement in the standards process. I think
the simplest solution would be to create a DMTF community committee
which may include various related cloud community groups such as the
CCIF, OCCI, Open Web Foundation, Open Cloud Consortium, Apache
Foundation and others. Each group cloud include a couple key community
leaders who would be authorized to sign the required NDA's / legal
agreements while also sharing any non proprietary details with their
larger community for discussion / ratification. I agree with Alex, the
Java Community Process (JCP) and Java Specification Agreement (JSPA)
would be great models to follow. (JSPA is an agreement between a
company, organization or individual, and Sun, setting out each
Community Member's rights and obligations when participating on the
development of Java technology specifications in the JCP.)

As for what we should do with the CCIF, at this point I would be in
favor of "donating" it to some other larger organization to help kick
start a community.

Mark, what are your thoughts?

Reuven

[Mark A. Carlson]

The idea of having a broader community associated with the DMTF effort
does have some merit IMHO.
It could help formulate requirements, provide coordinated feedback on
works in progress and help promote
the adoption of the standards.

The SNIA is trying this already with the SNIA Cloud
http://groups.google.com/group/snia-cloud/subscribe
Google Group that complements the work going on among SNIA members in
the Cloud Storage TWG.

DMTF already has a process for providing feedback:
http://www.dmtf.org/standards/feedback/
so no additional legal agreements need be made for community members to
provide their input.

-- mark

[Sam Johnston]

David,

Your feedback makes a lot of sense. Once upon a time standards were expensive documents complete with expensive royalties defined by way of expensive votes behind closed doors. Open standards changed this somewhat by being available without charge (both for access and use), but for many an "open standard" is simply that - any company can create an "open standard" by this definition, with or without community involvement (OOXML for example).

Some (such as DIGISTAN) are more focused on the open process, claiming that standards need to be "immune to vendor capture at all stages in its life-cycle", thereby making it possible to "freely use, improve upon, trust, and extend a standard over time". That is certainly the direction we've been going in for some time and while I think we'll get there eventually I'm somewhat more objective about it and am actively seeking a balance for the Open Cloud Initiative.

Correct me where I'm wrong, but my understanding is that the result of the DMTF effort will be heavily based on (if not simply a rubber stamping of) VMware's contribution of the vCloud API. Regardless of how elegant the result is, it gives VMware a huge advantage over their competitors while invariably shifting the center of gravity of the resulting standards towards their products. Of course most of the companies listed in the DMTF release are playing catch up (VMware having a multi-year jump start on them in the virtualisation space) and essentially have little choice but to support anything which gets them on the playing field.

That's a clever play for all involved, but it's not necessarily great news for end users (like my large enterprise clients) who are conspicuously absent from all of this. I'm happy to get involved and have emailed DMTF about the process but I wonder about the requirement/justification for this "sponsorship" step and how "open" this really is. It does however seem like a useful step in the ongoing evolution of virtualisation, if not necessarily game changing for cloud computing itself.

Sam

[Liming]

After having a look at the Review Board and Participation
Requirements, I am sure that this a game for big companies, as other
standards, this has nothing wrong, and other standards have done or
will be done like this.
My confusation is: where is the end user?where is the community?where
is the open source?

On 4月28日, 上午2时13分, Sam Johnston <s...@samj.net> wrote:
> David,
>
> Your feedback makes a lot of sense. Once upon a time standards were
> expensive documents complete with expensive royalties defined by way of
> expensive votes behind closed doors. Open standards changed this somewhat by
> being available without charge (both for access and use), but for many an
> "open standard" is simply that - any company can create an "open standard"
> by this definition, with or without community involvement (OOXML for
> example).
>

> Some (such as DIGISTAN <http://www.digistan.org/open-standard:definition>)

> are more focused on the open process, claiming that standards need to be
> "immune to vendor capture at all stages in its life-cycle", thereby making
> it possible to "freely use, improve upon, trust, and extend a standard over
> time". That is certainly the direction we've been going in for some time and
> while I think we'll get there eventually I'm somewhat more objective about
> it and am actively seeking a balance for the Open Cloud

> Initiative<http://www.opencloudinitiative.org/>
> .

>
> Correct me where I'm wrong, but my understanding is that the result of the
> DMTF effort will be heavily based on (if not simply a rubber stamping of)
> VMware's contribution of the vCloud

> API<http://www.networkworld.com/news/2009/022509-vmware-hails-new-vcloud....>.
> Regardless of how elegant the result is, it gives VMware a *huge* advantage

> > ....wanting to put actual engineers and time and money into trying to further

> > the interoperability and standards around hypervisors/virtual machine
> > technology at the file formats level, we are working with good intentions in
> > DMTF, I can assure you it is not for a bunch of sneaky reasons, it is
> > because it is the right thing to do.
>
> > We will do the same in SNIA around storage efforts, and if OGF OCCI has
> > some interesting API progress and gains some other significant members, well
> > OGF is also a terrific class-A SDO, we'll put some contribution there too.
>
> > Please support the DMTF effort. If you want to participate, and can't be a
> > full member, eg you need a sponsor to apply for the $200 individual
> > membership, I am sure any of the member companies including mine, will be
> > happy to oblige.
>
> > David Bernstein
>
> > Cisco.
>

> > *From:* cloud...@googlegroups.com [mailto:cloud...@googlegroups.com] *On
> > Behalf Of *Alex Esterkin
> > *Sent:* Monday, April 27, 2009 8:09 AM
> > *To:* cloud...@googlegroups.com
> > *Subject:* Re: DMTF to Develop Standards for Managing a Cloud Computing

> > On Mon, Apr 27, 2009 at 8:42 AM, Mark A. Carlson <Mark.Carl...@sun.com>

> > wrote:
> > > Open Cloud Standards Incubator Created to Address Management
> > > Interoperability for
> > > Cloud Systems
>

> > > PORTLAND, Ore. - April 27, 2009 - The Distributed Management Task Force

> > > -- mark- 隐藏被引用文字 -
>
> - 显示引用的文字 -

[Steve Loughran]

On Apr 27, 4:45 pm, Reuven Cohen <r...@enomaly.com> wrote:
> I think
> the simplest solution would be to create a DMTF community committee
> which may include various related cloud community groups  such as the
> CCIF, OCCI, Open Web Foundation, Open Cloud Consortium, Apache
> Foundation and others.

Speaking informally on behalf of the Apache Software Foundation, and
as one who has participated in standards in the past

* I see little value in the ASF participating in a community outreach
program where we lack voting rights or other control over the
direction of the process.
* Any process that requires NDA signing is something we are strongly
opposed to, as not only is it a barrier to entry, it is a barrier to
execution. You cant discuss it in open lists, in code, in bugreps,
hence it destroys the community.
* Any process which requires flying to meeting rooms round the world
is a major barrier to open source project participation.
* Any standard written without test cases is doomed to be an
interoperability nightmare and should not be something the ASF should
endorse. I would cite WS-Addressing as a case in point, stuff built on
top of it (WSRF?) as a consequence.

The testing point is critical; I have had to use standards that never
thought about testing; I have help specify things that turned out to
be untestable. I have also had to work in standards groups where the
attempts by some of us to define a test suite was viewed as holding up
the standardisation process. It does, if your goal is a piece of paper
that isn't implementable or interoperable. By thinking of testing from
the outset, you end up with better specs.

What the ASF is interested in is
1. stable cluster management APIS. EC2 is limited but works, Sun's
REST API probably adequate for most other needs. What value is there
in having more?

2. cluster time on different datacentres. This includes the storage of
large (open) test datasets as well as CPU hours.
We also need access to the test URLS, the ones that reject your
requests as you have asked for two many machines, etc. All
implementations test services should be visible too.

3. Defining and implementing the application architecture for the next-
generation cloud hosted applications. We don't need a JCP, it only
leads to the old politics.
https://svn.apache.org/repos/asf/labs/clouds/src/doc/apache_cloud_computing_edition_oxford.odp
This is what we want datacentre time for, and why those people
implementing it should work with the ASF: if we don't get the cluster
time, your infrastructure won't get supported.

If people are willing to help with this, we have an apache "lab"
project, Apache Clouds, ready for you code, tests and ideas.
http://labs.apache.org/

>Each group cloud include a couple key community
> leaders who would be authorized to sign the required NDA's / legal
> agreements while also sharing any non proprietary details with their
> larger community for discussion / ratification. I agree with Alex, the
> Java Community Process (JCP) and Java Specification Agreement (JSPA)
> would be great models to follow. (JSPA is an agreement between a
> company, organization or individual, and Sun, setting out each
> Community Member's rights and obligations when participating on the
> development of Java technology specifications in the JCP.)

Rights and obligations which Sun then ignored when refusing to release
the test kit (TCK) for Java to the Apache Software Foundation, so
preventing the ASF from validating their implementation of Java, and
hence automatically acquiring the group intellectual property rights
related to the implementation.
http://www.apache.org/jcp/sunopenletter.html

-Steve

[jeff wheeler]

Sam
I'd like to respond to the thread re: VMWare having both an upper hand
and the leading contribution to the DMTF efforts.
I just looked at the roster for the SVPC Working Group alone in the
DMTF (System, Virtualzation, Partitioning and Clustering standards)
and see 10 pages of corporate members.
The litany covers pretty much anyone with an active interest in
Virtualization and OVF including Microsoft, IBM, Cisco, HP, Nortel,
Harris etc...
Much work is done not just on the OVF but on the Virtualization Model
and associated profiles (implementation specific uses of the CIM
Virtualization model) as well.
VMWare does not have a leg up (so to speak) as OVF is truly vendor
agnostic. Multiple companies' strategies have contributed to, and
continue to do so, the OVF and Virtualization model maturity.
I'd encourage anyone who is seriously interested to pull down the
public docs on the Virtualization Model and the OVF from the DMTF
site, give it a read, and if there is interest join a call...
www.dmtf.org
regards
jeff wheeler

On Apr 27, 11:13 am, Sam Johnston <s...@samj.net> wrote:
> David,
>
> Your feedback makes a lot of sense. Once upon a time standards were
> expensive documents complete with expensive royalties defined by way of
> expensive votes behind closed doors. Open standards changed this somewhat by
> being available without charge (both for access and use), but for many an
> "open standard" is simply that - any company can create an "open standard"
> by this definition, with or without community involvement (OOXML for
> example).
>

> Some (such as DIGISTAN <http://www.digistan.org/open-standard:definition>)

> are more focused on the open process, claiming that standards need to be
> "immune to vendor capture at all stages in its life-cycle", thereby making
> it possible to "freely use, improve upon, trust, and extend a standard over
> time". That is certainly the direction we've been going in for some time and
> while I think we'll get there eventually I'm somewhat more objective about
> it and am actively seeking a balance for the Open Cloud

> Initiative<http://www.opencloudinitiative.org/>
> .

>
> Correct me where I'm wrong, but my understanding is that the result of the
> DMTF effort will be heavily based on (if not simply a rubber stamping of)
> VMware's contribution of the vCloud

> API<http://www.networkworld.com/news/2009/022509-vmware-hails-new-vcloud....>.
> Regardless of how elegant the result is, it gives VMware a *huge* advantage

> over their competitors while invariably shifting the center of gravity of
> the resulting standards towards their products. Of course most of the
> companies listed in the DMTF release are playing catch up (VMware having a
> multi-year jump start on them in the virtualisation space) and essentially
> have little choice but to support anything which gets them on the playing
> field.
>
> That's a clever play for all involved, but it's not necessarily great news
> for end users (like my large enterprise clients) who are conspicuously
> absent from all of this. I'm happy to get involved and have emailed DMTF
> about the process but I wonder about the requirement/justification for this
> "sponsorship" step and how "open" this really is. It does however seem like
> a useful step in the ongoing evolution of virtualisation, if not necessarily
> game changing for cloud computing itself.
>
> Sam
>
> On Mon, Apr 27, 2009 at 5:38 PM, David Bernstein (daberns) <
>

> > *From:* cloud...@googlegroups.com [mailto:cloud...@googlegroups.com] *On
> > Behalf Of *Alex Esterkin
> > *Sent:* Monday, April 27, 2009 8:09 AM
> > *To:* cloud...@googlegroups.com

> > *Subject:* Re: DMTF to Develop Standards for Managing a Cloud Computing

> > On Mon, Apr 27, 2009 at 8:42 AM, Mark A. Carlson <Mark.Carl...@sun.com>