DEWS Technical Workshop, 6-7 February 2007

Notes from the workshop

Aims of the workshop

The focus of the workshop will be integrating all the pieces of the DEWS system. The final goal will be to have end-to-end connection between the client systems (the DEWS portal and BMT SEAInfo) and the underlying data.

For each stream (marine and health) we need to connect the following components together:

1. [The incoming data feeds and the data store (I think this is already done)]
2. The Web Services layer and the data store (implemented in marine stream but not health)
3. The Gatekeeper (security layer) and the Web [Coverage/Feature] Services layer
4. The Gatekeeper and the auditing/logging system
5. The Web Portal and the Gatekeeper
6. The Session Manager and the Attribute Authorities (the AAs for health and marine are separate)
7. The Web Portal and the Session Manager
8. (marine stream only) The BMT system and the Gatekeeper
9. (marine stream only) The BMT system and the marine Attribute Authority

Please let me know asap if there is anything I've missed or misunderstood.

This workshop is all about interfaces so if there are any components that have yet to be created, they will be "stubbed out" (i.e. faked up) so that the real components can slot into the system later. If there is time, some of these components could even be created in the workshop.

We have two rooms available (except from 4:00 onwards on day 1 when we will only have one), so there is scope for parallel work. There will be wireless internet connectivity.

Location

ESSC, University of Reading: see http://www.nerc-essc.ac.uk/AboutESSC/Travel.htm for directions

Participants

• ESSC: JonBlower, AditSantokhee
• BADC: Phil Kershaw
• Lost Wax: Malcolm Smith, Brian Etheridge, Neil Stamps
• BMT: Mark Lidstone (1st day) Chris Rawlings (2nd day)
• IBM: Andrew Barnes (both days), Chris Whyley (second day only)
• Met Office: Chris Wood (both days), Jeremy Tandy (first day only), Liz Osborne (second day only), Bruce Wright and Lindsey Payne (both first day only, 4:30 onwards)

Agenda

It is my impression that most of the components of the DEWS system are near completion but I'm not sure about the components related to security. I anticipate that much of the workshop will consist of integrating the security components with the rest of the system and ensuring that they interoperate, especially from Python to Java/Websphere and vice versa.

First day: 6th Feb

Aim of the first day: To have a skeleton integrated DEWS system with all components represented but replaced with simple stubs where necessary. It is less important to integrate BMT SEAInfo today.

• 10:00 - 10:30: Welcome and coffee
• 10:30 - 11:00: Overview and discussion of agenda (Jon Blower)
• 11:00 - 12:30: Discussion of components in turn: For each component we need to clarify its high-level operations (i.e. its external interface), its status (is it finished? not yet started?) and its physical location: around 10 minutes each
  • The Portal: how it communicates with the rest of the system (Brian E)
  • The Gatekeeper Web Service (Adit Santokhee/Brian Etheridge)
  • The Session Manager Web Service (Phil Kershaw)
  • The Attribute Authorities (Phil Kershaw)
  • The Auditing and Logging Web Services (Brian Etheridge/Chris Wood)
  • The Marine Web Coverage Service (Adit Santokhee)
  • The Health Web Feature Service (Chris Wood)

For example, for the Marine WCS, its high-level operations are getCapabilities(), describeCoverage() and getCoverage() (it also has operations for checking the status of an ongoing data extraction). Its status is "ready" (so we can test the live system) and its physical location is ESSC.

12:30 - 1:30 LUNCH

• All afternoon: Connecting components together, using stubs if the components do not exist. Aim to have a complete integrated skeleton system, perhaps minus BMT SEAInfo.

Other important afternoon items:

3:00 COFFEE

4:30-5:30: Adit, Neil, Bruce and Lindsey to discuss marine databases

I also note that a couple of "side-meetings" have been requested on this day, one on problems with describeCoverage() in WCS and one on discussing the Met Office's WCS (created by a summer intern) and the DEWS WCS. I suggest that we fit these in if we can but these discussions could also be held in a different meeting between ESSC and the Met Office: they don't require everyone's presence.

Second day: 7th Feb

Aims of the second day: To replace the stubs in the integrated system one by one with real components. To integrate BMT SEAInfo.

Hard to plan a detailed agenda in advance for this day. I suggest we let the experiences of day 1 drive what we do today. The most important thing is to end up with an integrated DEWS system, using real components where possible, and that everyone understands. Any remaining stubs can be replaced with real components off-line if we run out of time in the workshop.

Chris Whyley of IBM will hopefully be able to attend so he would be able to help with Websphere issues.

9:30 Start

10:30 COFFEE

12:30 - 1:30 LUNCH

3:00 COFFEE

I don't know when everyone needs to leave to catch trains etc but we will probably finish around 4:30-5:00 if that suits everyone.

Original suggestions

Phil Kershaw:
    * the gatekeeper
    * BMT SeaInfo client to GADS-WCS and polling of gatekeeper for responses
    * The audit / logging web service
    * The security installations at LostWax and ESSC
    * requests wireless connection

Brian Etheridge:
    * Marine web services interface
    * Health web services interface
    * Session Manager interface
    * Gatekeeper interface
    * Gatekeeper internal processing 

Jeremy Tandy:
    * Side-meeting about WCS/WFS and problems with describeCoverage

Adit:
    * Side-meeting about using Met Office's n-D WCS from intern's work

Keith:
    * Knowledge transfer with Grid Computing Now!

Malcolm:
    * End-to-end DEWS portal functionality

Chris Wood:
    * Gatekeeper interaction (ie. trying to get it working)
    * Installation and configuation of the Met Office Attribute Authority. It is very
unlikely that I will be able to get access to our internal systems from Reading but
if we can get it up and running on my notebook it should be easy enough to
duplicate when I get back.
    * Demonstrating the health Geoserver WFS.

-- JonBlower - 31 Jan 2007