Re: [ggf-ogsa-sec-wg] VO consideration

["Takashi Kojo" <tkojo@mvi.biglobe.ne.jp>]

Glenn,

Thank you for your comments.

> > My motive is to find out if/how the VO model is sufficient enough
> > to deal with real use cases of the grid regarding with grid security.
>
> VOs, as you point out, do seem to mean different things to
> different people. In many grid installations, a set of directories
> cataloging the VOs users and resources, a CA which issues certificates
> that the VO resources accept and appropriate local (to the individual
> resources) security mechanisms constitute a VO.

In another word, VO is a virtualized CA domain.

> At the University of Virginia, we've been doing research on VOs
> and are designing policy languages and mechanism to more easily create
> virtual organizations. The main idea is to support formal, virtual
> organization-wide policy that allows for autonomic enforcement/management
> of that policy.

I would appreciate for pointers to the resources.

> > - Would a single layer of VO be sufficient enough?
> >     to deal with levels of trusted users or VOs...
>
> Certainly this is a needed capability. The forth-coming
> specifications WS-Trust and WS-Policy will be important here.

Are they the efforts Phil mentioned?

> > - How do you set up/modify the VOs with which authority?
> >          or maybe can you partially define them statically, too?
> >       What would be a set of services of VO manipulations?
>
> Of course, this depends on how you define your virtual
> organization. Setting up and running a CA (if you require this) can be a
> complex operation. The goal of our research is to allow a VO creator to
> specify VO policy and have a set of tools that generate the necessary
> mechanism (e.g. grid services) that operate the VO according to that
> policy.

I assume this is about the diagram on the roadmap.


> > - What would be real organization(RO) policy/trust and VO policy/trust?
>
> Again, WS-Policy and WS-Trust will play a role here. Policy
> conflict between VO-wide policy and the policies of the various ROs in the
> VO is an area of active research for us. Policy and trust are both dynamic
> considerations and so there must be mechanisms to handle updates.

Actually I am not yet very clear about the policy and trust relationship.
In my sense, policy might include non-security related policy
that could  be defined for administration or system efficiency, for example.
But let's assume, for now, WS-Policy is limited to security related policy.
Then I agree that each RO should have its trust model and claim
of the own trust. Looking at a party's trust claim and own policy
you can translate them into a set of operation rules. On the way,
you might also find some collisions of the trust and policy.

Is this correct interpretation?


Kojo