Let me see if I can clarify:

Airavata expects clients (or gateways) to manage users and groups. And Airavata 
relies on gateway portal request (which it will blindly trust-based on well 
secure communications) to tell which user and group is making a request. The 
discussion I see now happening is for Airavata to manage the input and output 
data. So it will need its own internal structure on how to manage these. 

If a gateway choose to, it can leverage this structure, but this will not be 
required. 

As for the data model:

I suggest against making any implicit derivations of the data models these 
leads to ambiguity once the motivation is forgotten. I rather suggest making 
explicit changes to the data model if the need is justified.  

Airavata data model as I understand is as follows:

Gateways-> Users-> Projects->Experiments

I think what we are talking is to change it induce a groups in between which is 
a fine addition, so then it should be:

Gateways-> Groups-> Users-> Projects-> Experiments.

Simpler use cases may choose to always use a single default group and a user 
may choose to have single project. In that case, it shows like:
Gateways->Users->Experiments, but in the backend its always should have a 
self-explanatory hierarchy. 

Suresh


On Jan 31, 2014, at 2:49 PM, Miller, Mark <[email protected]> wrote:

> I think we agreed that the Gateways would be responsible for user management, 
> but maybe that is at a different level?
>  
> From: Sachith Withana [mailto:[email protected]] 
> Sent: Friday, January 31, 2014 11:46 AM
> To: [email protected]
> Subject: Re: Users and UserGroups in Science Gateways
>  
> Thanks Raman,
> I looked into the tables. It looks to be a viable option.
> But is it okay to manage the users from the Airavata server itself?
>  
> 
> On Fri, Jan 31, 2014 at 1:03 PM, Raminder Singh <[email protected]> 
> wrote:
> Looking at Airavata Data model, there is project which is equivalent to User 
> group. A project can have multiple user for a gateway and a gateway can have 
> multiple projects.
>  
> Thanks
> Raminder
>  
> On Jan 31, 2014, at 12:44 PM, Saminda Wijeratne <[email protected]> wrote:
> 
> 
>  
>  
> 
> On Fri, Jan 31, 2014 at 9:28 AM, Miller, Mark <[email protected]> wrote:
> As an end user, I think even with just a single tool, the argument for a 
> group is there.
> As Saminda suggested, one may wish to have a join area to work in with a 
> group, and individual work you aren’t ready to share, even with just a single 
> tool.
>  
> Because a Gateway is wide open, it is easy to imagine a case where there are 
> several groups, each with (potentially overlapping) users, who all are 
> working with Amber for different projects..
> Agreed. I totally forgot about shared projects scenario. Thanks for pointing 
> it out Mark. 
>  
> Mark
>  
> From: Saminda Wijeratne [mailto:[email protected]] 
> Sent: Friday, January 31, 2014 9:11 AM
> To: dev
> Subject: Re: Users and UserGroups in Science Gateways
>  
> I'm not an expert on the Amber usecase, but I suppose it is possible to have 
> multiple user groups for a single usecase. It could be to have a shared space 
> of data hidden from each user group. It could be to author 
> access/privilledges to the portal or resources depending on the user group.
> 
> I would assume the mapping of user to usergroups would be many to one 
> mapping. Unlike user roles I don't think it is sensible to have users being 
> mapped to multiple groups (there is a whole area of security to consider 
> here).
>  
> (I do have to note that usergroup IMO is not a must concept for a science 
> gateway)
>  
> 
> On Fri, Jan 31, 2014 at 8:55 AM, Sachith Withana <[email protected]> wrote:
> Hi all,
>  
> Can someone explain to me the mapping of users to groups in a Science 
> Gateway? ( This is regarding the Amber usecase).
>  
> Can there be multiple user groups in a gateway portal when it comes to just 
> one usecase?
> ex: If we have only one usecase for the portal ( such as Amber)
>  
> --
> Thanks,
> Sachith Withana
> 
>  
>  
>  
> 
> 
>  
> --
> Thanks,
> Sachith Withana
> 

Reply via email to