CAIGEE

1
CAIGEE

• CMS Analysis an Interactive Grid Enabled
  Environment

2
What is CAIGEE?

• Emphasis on interactive vs. batch
• Low latency
• More unpredictable behaviour
• not exclusive to interactive analysis
• An integration of existing tools into a larger
  grid analysis environment
• Concentrates on CMS analysis, but could be
  applicable in other grid environments.

3
Requirements

• Build on the existing CS11 requirements and
  HEPCAL use cases
• Work with ARDA, CS11, and other Grid analysis
  efforts as much as possible
• Support both interactive and batch analysis
• Low latency and short running interactive jobs
  vs. higher latency long running batch jobs.
• Use language and OS neutral APIs
• Fault tolerance
• No single point of failure
• Adaptability
• Exploit usage patterns to optimize overall system
  performance
• Scalability
• Security
• Authentication, Authorization
• Support for heavyweight and lightweight clients
  (web browsers, handheld devices)
• Adapt existing tools wherever possible

4
System features

• Hierarchical Peer to Peer architecture
• Better able to handle single node failure
• Dynamic host discovery
• Distributed lookup service
• Web Services APIs
• Language and OS neutral
• Multiple service instances accessed through a
  distributed lookup service
• Failed service instances can be bypassed by
  locating a new instance

5
Standards

• Describe Web Services using WSDL
• Provide SOAP and XML-RPC access to services
• X509 based security through HTTP, HTTPS, and GSI
• OGSA compatible services

6
Analysis Client
Classes of GAE Services
Analysis Web Clients
High-Level Web-service Grid Services
Request Servicing
Data Discovery
VO Structure
Application Management
VO Management Service
Metadata Service
Monitoring Service
Process Wrapper Service
Lookup Service
Virtual Data Service
Request Steering Service
Request Scheduler Service
Replica Location Service
Request Execution Service
High-Level Web-service Grid Services
Middleware Grid Services
Grid Resource
Data Service
VO Mngmt. Service
Replica Loc. Service
Data Transfer
Gatekeeper
Monitoring Service
7
CAIGEE Architecture
Web Client
Web Client
Web Client
Grid Services Web Server
Scheduler
Catalogs
Fully-Abstract Planner
Metadata
Partially- Abstract Planner
Virtual Data
Data Management
Grid-enabled Domain Specific Codes/Applications
Monitoring
Replica
Fully- Concrete Planner
Execution Priority Manager
Grid Wide Execution Service
8
CAIGEE Architecture II
9
Service Flow
10
Analysis Client
Classes of GAE Services
ROOT- Clarens
Analysis Web Clients
High-Level Web-service Grid Services
Request Servicing
Data Discovery
VO Structure
Application Management
VO Management Service
Metadata Service
Monitoring Service
Process Wrapper Service
MonALISA
Clarens
Lookup Service
Virtual Data Service
Request Steering Service
Chimera
Clarens
Request Scheduler Service
Replica Location Service
Sphinx
RLS
Request Execution Service
VDT-Client
High-Level Web-service Grid Services
Middleware Grid Services
Grid Resource
Data Service
VO Mngmt. Service
Replica Loc. Service
Data Transfer
Gatekeeper
Monitoring Service
VDT-Server
Clarens
Clarens
11
Future Work

• Add asynchronicity to services
• Investigate and prototype JXTA as a peer to peer
  protocol
• Write generic service APIs for existing service
  implementations
• Implement a prototype system based on the SC2003
  demonstration
• Add more services to prototype system as they
  become available
• Add OGSA support to services

12
GAE Collaboration DesktopExample

• Four-screen Analysis Desktop 4 Flat Panels 6400
  X 1200
• Driven by a single server and two graphics cards
• Allows simultaneous work on
• Traditional analysis tools (e.g. ROOT)
• Software development
• Event displays
• MonALISA monitoring displays Other Grid Views
• Job-progress Views
• Persistent collaboration (VRVS shared windows)
• Online event or detector monitoring
• Web browsing, email

13
Grid-Enabled Analysis Prototypes
Collaboration Analysis Desktop
COJAC (via Web Services)