OGSA Interim Meeting #11 - Day 2 (May 24, 2005) =============================================== * Participants Jay Unger (IBM) Jem Treadwell (HP) Andreas Savva (Fujitsu) Takuya Mori (NEC, ANL) Mark Morgan (UVa) Steve McGough (LeSC) Fred Maciel (Hitachi) Hiro Kishimoto (Fujitsu) Andrew Grimshaw (UVa) Donal Fellows (Uni. of Manchester) Abdeslem Djaoui (RL) Mathias Dalheimer (Fraunhofer-ITWM) Dave Berry (NeSC) On bridge: Tom Maguire (IBM) Action Summary: (See body of minutes for details.) [Roadmap document] Action: Dave B will draft text for the compliance section to replace existing text. Action: Andreas and Jem to complete Architecture and Glossary 1.5 entry. Action: Andrew to check whether BP will be used by RNS and to also arrange for a revision of the RNS section. Action: Abdeslem to provide an entry for a Basic information monitoring profile Action: Jem to move Naming section to section 4. Action: Jem to delete the OGSA Basic Data profile section Action: Jem to create a new section for "Future documents/profiles" and put information for the EM and Data profiles there. Action: Dave B to write up an entry for a possible future Data profile. Model it along the lines of the EM profile entry. Action: Takuya to write an entry for a future Security profile. Model it along the lines of the EM profile entry. Action: Jem to delete the "Service descriptions and scenarios section" Action: Andrew and Mark to write up a paragraph for UVa OSS OGSA activities. Action: Hiro to write up a paragraph for Business Grid OSS OGSA activities. Action: Kazushige Saga to write up a paragraph for NAREGI OSS OGSA activities. Action: Jem to re-organize the Authors/Editors and Contributors/Acknowledgements as discussed. Action: Andreas to create a post-v1 Roadmap tracker and move deferred items there. [OGSA-WG Management] Action: Hiro to announce Tom Maguire as the group's choice for new co-chair on the list. (Done.) * Recap of Day 1. See minutes of Day 1. * Roadmap document review https://forge.gridforum.org/projects/ogsa-wg/document/draft-ggf-ogsa-roadmap/en/8 Review contacted first by going through document and reviewing comments or incomplete sections. And then by reviewing the tracker artifacts. ** Document/Comments review *** Compliance tests statement and Compliance section - Compliance test may be future work, perhaps by someone else. There is a (currently empty) section on compliance in the document. - Agreed that it is important to make a clear statement in the roadmap that we do not do this and hence there is no way to claim "OGSA compliance"; so preempt people who may want to make that statement without sufficient justification. And that we look for someone in the future to do it. - Deleted comment. Action: Dave B will draft text for the compliance section to replace existing text. *** HK2 comment: - The comment is from Dave S on OGSA leadership - Agreed to delete and review the relevant tracker *** Branding section: - This section is a placeholder. It is not within the OGSA-WG's purview to decide this. - Dave S is now writing the GGF roadmap document. It includes policy statements on branding and what OGSA profiles are. - Agreed to leave section as-is for now and replace with a summary from Dave S's doc including a reference to it when it is available. *** Architecture doc 1.5 and Glossary section - Need feedback from section owners first. - Action: Andreas and Jem to come up with schedule - (Done later in the day: Agreed to a first draft by July, public comment by GGF15 (Oct) and publication by January.) Action: Andreas and Jem to complete Architecture and Glossary 1.5 entry. *** Byte IO section - Corrected: March 2006 not Spring 2005 *** Data Architecture section - Corrected: June 2006 not Summer 2006 *** OGSA Naming requirements section - This is a new section added by Andrew G - There is a draft doc produced by the Naming design team - Agreed that title is "Naming issues in distributed systems" and changed. *** Byte IO Interface section - Corrected: Spring 2005 -> March 2006 *** ACS section - This is the revised version, the first draft of which was reviewed during a teleconf - Agreed that adding additional information about open source implementations in the body of the milestones section if fine as long as who is committed to such an implementation can be identified. *** BES section - Remove line from table about experience document - No need to list both WSRF and the 'WSRF BP' since the BP includes WSRF. - Deleted WSRF line. - What does the information in the tables mean? When is it assumed to be accurate? Consensus that the table should be filled in with information correct at the time of the publication of the roadmap. This implies that the **Editor of the Roadmap document** is responsible to check the overall accuracy of all entries. (See further down "Authors/Contributors" on Editor choice.) *** JSDL section - Corrected table entries/dates - Added statement that CIM and POSIX 3 are not required for a JSDL implementation. *** RNS section - Discussed milestones. It seems too soon to put Sep for GWD-RP. It should be changed to October. - Is WSRF used or not? (Or WSRF BP). - Not sure. It is probably used by the current draft but it is not clear that this is necessary. Action: Andrew to check whether BP will be used by RNS and to also arrange for a revision of the RNS section. *** OGSA Basic information monitoring profile Action: Abdeslem to provide an entry for a Basic information monitoring profile *** BP section---ok *** Naming section - Agreed that this section should be moved to section 4. It is a normative specification not a profile - Table: Change Spring to March 2006 - The status of WS-Addressing is Evolving (not finished yet) - The group's name is likely to be "OGSA Naming WG" Action: Jem to move Naming section under section 4. *** OGSA Basic Data Profile - No schedule for this so agreed to delete from roadmap. Action: Jem to delete the OGSA Basic Data profile section *** OGSA Basic Execution Management profile - Agreed that it is still too early to start thinking about this. Nonetheless it is important to keep some information in the Roadmap to show what we have in mind even if there is no concrete schedule. - Agreed to make a new section for Future profiles or documents that are envisaged as needed. - Agreed to put the EM profile and Data profile there. Action: Jem to create a new section for "Future documents/profiles" and put information for the EM and Data profiles there. Action: Dave B to write up an entry for a possible future Data profile. Model it along the lines of the EM profile entry. *** Security profile - Basic requirements are covered by the WSRF BP - Agreed that a more general security profile falls under the category of future profiles Action: Takuya to write an entry for a future Security profile. Model it along the lines of the EM profile entry. *** Informational profile - Nothing applicable. *** Service descriptions and scenarios - There are a type of informational documents. Agreed to delete the section and include any entries in earlier sections as appropriate. Action: Jem to delete the "Service descriptions and scenarios section" *** Standards landscape draft - Reviewed liaisons for outside organizations. Note that these are the OGSA-WG liaisons not the GGF ones. - Added Tom M as liaison to DMTF UC (sharing DMTF with Fred) *** SCRM section - Introduction of SCRM by Hiro, including motivation, history and plans at GGF14. - A standards landscape document is planned and the schedule is included. *** Open Source software - It is not correct to talk about globus consortium here. The Globus toolkit is open source software. The consortium on the other hand is developing or funding software that will be contributed back to the open source project. - Agreed to editing changes in various place from 'consortium' to 'project' - Rewrote section to talk about a number of projects in general and then make clear that Globus toolkit is one example - Also agreed to ask people doing related work to contribute a paragraph similar to the Globus toolkit one they should and it will be included in this section. Volunteers: - UVa paragraph by Andrew and Mark - Business Grid : Hiro - NAREGI : Kazushige Saga Action: Andrew and Mark to write up a paragraph for UVa OSS OGSA activities. Action: Hiro to write up a paragraph for Business Grid OSS OGSA activities. Action: Kazushige Saga to write up a paragraph for NaReGi OSS OGSA activities. *** Security section revision This section is required by GGF. *** Authors/Contributors etc - Andreas proposed Jem as the Editor of the Roadmap - Jem accepted as Editor. - Discussed how to format Authors/Editors/Contributors/Acknowledgements - Change Current Authors to Contributors and list just Editor(s) at the front. - People who submitted sections should be Contributors - Also list people taking part in the review Action: Jem to re-organize the Authors/Editors and Contributors/Acknowledgements as discussed. *** Tracker review - Did a tracker by tracker review and updated individual entries - For details see individual tracker artifacts https://forge.gridforum.org/tracker/index.php?func=browse&group_id=42&atid=789 - Deferred artifacts should be moved to a separate post-v1 tracker. Action: Andreas to create a post-v1 Roadmap tracker and move deferred items there. * OGSA-D status review A review for those members not actively involved in the OGSA Data activities. https://forge.gridforum.org/projects/ogsa-wg/document/ogsa-data-summary/en/1 ** Overview - Group was recently approved - Getting started with gridforge; adding members etc - Reviewed the Architecture document structure - A number of issues: security aspects especially wrt Federation. E.g., are these unique to data or not; or what is specific to data. - Is INFO-D included in the architecture or not? - Thinking of splitting INFO-D in two parts; one specific to the data architecture and one not. Nothing is decided; it is still under discussion ** Review of issues - Protocol: how the decision of which protocol to use for the transfer - Control: should there be a separate control entity to manage the transfer - And how heavy/light weight things are depending on the choice taken - Storage management (SM) - Open issue on how to get the various emerging specs to play well together - SM defines own artifacts: logical file name, and files : Can a single namespace be defined? (E.g., use RNS and Byte-IO) - SM may not even be using WS-Addressing - Coherency management - How to specify policies. - 'policy' may be QoX instead; That is a 'service level': not just performance but other aspects that come out as some kind of system quality. - Might be interesting to come out with an initial (textual) description of what the ranges of this area are. (A landscape of sorts.) - Security - Data specific security issues: Two main categories: - persistent copies - on wire - If a replica is made it must enforce the same policy even when that policy changes at the 'original'. In other words there is a coherence issue on policy. - And different organizations may have different needs on how fast to make the policy coherent - And therefore may have different tradeoffs - Security of backups - Limit on the number of queries (more of a privacy issue) - What is the state of the OGSA Security design team state and can it help here? - Some activity is on-goining but the general Security area does not seem to be very active. - Transactions - Not a lot of work going on. Hoping that it would be possible to layer them later. - There are a number of specs possibly related to this functionality in the WS space. They should probably be reviewed. ** Issues from OGSA F2F: - Resource framework, description language, naming, etc - Scheduling - Information should be provided to schedulers - A high level characterization of what kind of attributes are needed in this space is needed as a first step. - Some OGSA-WG members are thinking of doing some work in this area but the main focus is BES and JSDL. - Relation between Data and other parts of OGSA? - Data is a resource (one facet) - Like other resources it has scheduling attributes; some of these may be unique to Data; some not. - Should identify these attributes as far as possible and in comparison to ones for other kinds of resources * OGSA documents schedule Final Proof GGF-Editor (On hook) - BP 1.0 5/27 6/1 6/3 (Tom) - Profile def 1.0 6/3 6/6 6/10 (Dave S -> Tom) - Roadmap 1.0 6/3 6/6 6/10 (Jem and Hiro) Andrew is the AD and will be reviewing these docs before they reach the GGF Editor. Since he is already familiar with them it is expected that they will move to the GGF Editor soon after submission. * GGF14 schedule review - Review of current schedule proposal. - Confirmed owners, agenda and so on - Information session: One line agenda: Architecture and related specifications * Naming WG review - Started regular calls - Charter draft revised and sent to Steve Crumb - Proposed co-chairs: Andrew & Manuel Pereira - Was hoping for an MS co-chair but it did not work out. - The plan is to take over the RNS spec once it is put in the GGF process; follow up for public comment and so on. - The spec includes a resolution service and has a GFS appendix, which does not seem appropriate. Probably need to make the GFS specific part a separate document - WS-Naming document - This draft was also reviewed by the MS Indigo team; received feedback. - Why is not WS-Addressing and WS-Discovery sufficient for resolution. - Tooling doesn't use To: field as URI but as URL; WS-Naming to work around tht problem - WS-Discovery geared to solve a different problem and is more heavy-weight - Andrew and MS contact to write rebuttal of received statement - WS-Naming resolution document - OASIS XRI specification review - This seems to be more closely related to RNS - Solves the 'what's the root?' problem. Can specify with special character which root to use. - They don't operate with EPRs - In terms of properties they could be one instance of an abstract name; in the sense that an abstract name is not constrained to any specific structure. - Handle.net - Licensing reviewed; Commercial use is not free - Also client software must be aware that it is dealing with handle.net entities - WS-Naming is an extension to EPRs and can be safely ignored by client software. (An EPR with such an extension can still be used as a normal EPR.) - Can RNS be used to name 'small' pieces of data, e.g., a database record? - Naming 'small' stuff involves generating an EPR. It is a string operation and so it is not a heavy one. - An 'entity' does not have to be a 'resource' to be nameable by RNS. (This is probably something that needs more clarification especially in terms of terminology.) - Also need review/update of OGSA Glossary. - Review of abstract name, requirements and positioning in EPRs - Not a panacea; an extension so that some extra properties can be provided. - Discussion points: - Is it a 'normative' resolver in the EPR or not - Is the resolver entries 0-1 or 0-n - These are probably best left to the Naming WG. * OGSA-WG Management - Ian Foster's recent resignation has left Hiro as sole chair - Discussed requirements for choosing co-chair - Requires approval of GFSG - Proven ability to deliver - Proven technical skills - Proven ability to manage the group - As well as the rough consensus of the group - Should it be from academia or industry? - Consensus that it is more important to choose the right person - A number of potential candidates were named and discussed. Many declined the offer for their own reasons. - Tom Maguire was the group's unanimous choice - Tom accepted pending management approval. Action: Hiro to announce Tom Maguire as the group's choice for new co-chair on the list. (Done.)