[ogsa-d-wg] Telcon Agenda, 16th November - Storage management section review

Peter Kunszt peter.kunszt at cern.ch
Sun Nov 27 16:16:19 CST 2005


hi all


please find attached a new storage chapter version (chapter 7).
i've tried to work in the comments by allen and stephen. some
comments/points to discuss further:


- comment by allen on tying this chapter 'back to the QoS' discussion
earlier in the doc - i could not find the corresponding place. in the
versions i was looking at QoS was only mentioned marginally. which
passage did you refer to?

- i removed the reference to encryption. it is true that encryption
algorithms, key management etc can be considered an external service to
the data storage resource. it is up to the client to put encrypted data
into the storage space, not necessarily up to the storage to provide the
facility (which is a nice way to dodge this difficult topic :-) i hope
you agree.

- I decided to also remove the 'space partition' property. it was too
complicated and the same can be achieved by a set of spaces provided by
the storage instead. if the user requests a space which is too large,
the storage resource may simply negotiate a smaller one, and the user
may ask for more of those. this pushes the management of a set of spaces
to the client which is probably the right thing to do (SOA). comments?

- about security semantics of file space types (ie ones having a file
namespace): allen asks whether this should be defined here or by the
security group. i would argue that it should be defined here as they are
tied to storage security semantics. the ogsa security groups deal with
generic security aspects and semantics for authentication authorization
and accounting. but it is up to each specific service to provide the
corresponding semantics using these tokens. for example in a file space
the principals provided through ws-security (authz) can be used to set
and check the ACLs. so in summary: the security semantics of data
services are up to the data services to define. this is of course only
my opinion based on my current understanding and i'm happy to be
convinced otherwise.

- costing of storage. to define the cost of a resource sounds like a
generic problem which i believe should be discussed in a dedicated group
of the OGSA design. is this already done? i've simply put a forward
reference to it.

- i have (re-)added the chapters on security and transfer. the security
discussion is new and has the word discussion in its title. it contains
some additional aspects to consider that were not in the chapter yet.
the transfer chapter got dropped from the last version for no good
reason so i put it back.. if we decided to drop it for a reason, please
remind me.


sorry for the delay,

peter

ps. i'll try to do the use cases as well before the phone con, i got
very good input from stephen already.
 
-- 
------
CERN, 1211 Geneva 23, Switzerland

-------------- next part --------------
A non-text attachment was scrubbed...
Name: StorageChapterv2.doc
Type: application/msword
Size: 61952 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/ogsa-d-wg/attachments/20051127/c56386fd/attachment.doc 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 1395 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/ogsa-d-wg/attachments/20051127/c56386fd/attachment.bin 


More information about the ogsa-d-wg mailing list