060908-perfSONAR-Dev
From GEANT2-JRA1 Wiki
The objectives of this phone call are to discuss the topology measurement archive service, and if time perhaps discuss how (and when) others can start using the SQL MA. Additionally, we could also have a round of updates on what people are working on as it has been some time since we talked.
Information
Time: 0700 PDT, 1000 EDT, 1400 UTC, 1600 CEST, 1700 EEST
Gatekeeper: pgk.vc.dfn.de (for non-GDS members)
- GDS number: 004910091212314***4522
ISDN 0049 30 254 10800
- Listen to the lady
- Type conf-id 004910091212314***4522#
Backup - Skypecast: I did setup a skypecast backup. It is a little unclear how to get a direct URL to it from the skype web site (at least for a future skypecast). I set this up for 15 minutes before the meeting is actually supposed to start, so I will see if this URL actually works then. (And modify it here if needed.)
https://skypecasts.skype.com/skypecasts/skypecast/detailed.html?id_talk=32561
Agenda
- TOPS
- Overview of functionality - Ulisses
- Schema requirements
- Service design
- Question/Answer session
- General updates from development team
Minutes and Actions
Attending: Andreas, David, Ulisses, Martin, Jason, Jeff
Ulisses described his service from a use case and design perspective.
Jason pointed out the need to understand the expected usage of the service from a message request/response perspective. What messages are expected at what points for the expected usage patterns?
Jeff suggested that the 'trap' message generated from a topology database change should include the information that changed instead of just a key. This would immediately give the client the information they need saving the need to then request the data. Additionally, this would allow the same message format to be used for a direct query for data as for the 'trap' generated data simplifying the implementation. Ulisses agreed this was message flow optimization was desirable.
There was some concern by Ulisses that the SQL topology schema did not currently map very well with the XML topology schema. Martin asked Andreas to update him on the latest developments of the SQL topology schema so he could look at any discrepancies.
There was a fair amount of discussion about how the SQL topology would be mapped to XML in general. Specifically, there was discussion on how queries would be represented. Ulisses discussed using an XPath or XQuery syntax but only allowing specific queries that fit into a specific 'template' that could be mapped to an SQL query.
Actions:
- Ulisses will send out new messages request/response representing the use cases including examples of the query syntax.
- Call attendees will review messages. Another meeting will be called after everyone has reviewed this latest round of development.
- Andreas will make the CNM/Nemo SQL topology schema available somewhere the rest of the group can see it. (It is currently on the internal Geant wiki.)
