Transcript An architecture for remote processing

Remote Visualisation System
(RVS)
By: Anil Chandra
Visualisation Requirements

Obtain image data residing on different machines at
remote locations.

Visualise the images – whole images or subsets of larger
images.

Ability to interact with (analyse) visual data

Allow overlaying of catalogue information

Provide services at a satisfactory speed
Current Model 1
1. User requests an image/data from a particular service.
2. User receives a rendered raster image (e.g. gif) or
FITS image.
3. User visualises downloaded image on local computer.
Data query
User
Image
Database/Catalogue
of images
Current Models (cont)

The current model is an example of client side
visualisation.
- All processing is done on the client (user’s
computer).
What is RVS?

A distributed software system that allows scientists to
view and analyse astronomical data that are stored in
remote locations.

A server-side visualisation system
User command
Data request
RVS Server
Image + metadata
User
Raw image data
Database/Catalogue
of images
Client User Interface

Web browser or standalone application
RVS Server

Most processing occurs on the RVS server, not on
the client computer.

Obtains and stores images from
archives/databases

Processes images based on user commands

Uses AIPS++ Display Library (but not restricted to)

Provides its service via SOAP
RVS Server Design

Distributed architecture (via CORBA)
-
Scalability
-
Performance
-
Extensibility
Client Request
Web Server
Security Service
Data Centre Manager
AIPS++
Data Storage
Client – Server Communication

SOAP Interface to clients
-
Standard and popular protocol for web services
-
Allows custom clients
-
Based on XML, so it’s programming language and platform
independent
Why Server-Side Visualisation?

More processing power on servers than on the average
client PC

Servers can hold larger images than the average PC due
to more memory and storage space

Servers can receive larger images faster via higher
bandwidth connections to data centres

Any time, any where access
Server-Side Visualisation (cont)

Software maintenance/upgrades transparent to users

Only the server needs to be upgraded, not all the clients.

Existing tools such as AIPS++ can be utilised (no need to
install on clients)

Supports concurrent users access to same session
Disadvantages of Server-Side
Visualisation

Difficult to design/implement

Slower interactivity

Poorer performance with smaller size data that require
little processing

High reliance on server availability/integrity.
The VO Context

RVS is a visualisation tool amongst the many tools in the VO.

RVS may be used:

By end users to view and analyse images
In conjunction with other services (such as data mining)
By data centres to provide a visualisation interface to their data
It’s based on a more generic framework that may be used bring
other services, besides visualisation, online.
Project Status

High Level Design complete

A simple prototype (proof of concept) is in progress.
-
To be completed by 28 Nov 03

Project completion date: 25 May 04
-
A complete RVS Server
-
A demonstration client