CaGrid:ProductionGrid caGrid1.2

From caGridWiki

Jump to: navigation, search

Contents

caGrid 1.2 Production Grid

Overview

 caGrid 1.2 Production Grid
Enlarge
caGrid 1.2 Production Grid
  • caGrid itself provides no real “data” or “analysis” to caBIG™; its the enabling infrastructure which allows the community to do so Community members add value to the grid as applications, services, and processes (for example: shared workflows)
  • caGrid provides the necessary core services, APIs, and tooling
  • Community provided services:
    • Data Services
      • Expose data to the grid in a unified way
    • Analytical Services
      • Expose analytical operations to the grid
    • Local Deployments of Core Services (e.g. security services)
  • The real “value” of the grid comes from bringing this information to the “end user”
  • Community members develop end user applications which consume of the resources provided by the grid

Current Version

We are pleased to announce that the transition to caGrid 1.2 infrastructure has been completed. The caGrid 0.5 infrastructure and associated tools have been desupported as of January 1, 2008. All caGrid services and tools should use the caGrid 1.2 production infrastructure. You can see information here on how to request an account.

If you have any questions, please do not hesitate to request support.

Service Addresses

The design of caGrid is such that one needn't hardcode the addresses of services, as each running service registers itself to the Index Service. Registered services can then be discovered programmatically. In this sense, the Index Service is really the only service that a "well known address" is needed. However, in some cases it may be useful to know the address of core services. The NCI maintained production deployment of the core caGrid infrastructure services are:

The production grid can also be observed on the caGrid Portal.

Production Grid Hardware

Services for the NCI Production grid are hosted on multiple servers:

Physical Machines
OS CPU RAM Disk Services
RHEL4 2 x Intel 3.8 GHz 12 GB 2 x 146 GB, 2 x 72 GB Dorian, syncGTS
RHEL4 2 x Intel 3.8 GHz 12 GB 2 x 146 GB, 2 x 72 GB Authentication Service, CDS, syncGTS
RHEL4 4 x AMD 2.8 GHz Dual Core 64 GB VMWare ESX 3.02 VMHost #1
RHEL4 4 x AMD 2.8 GHz Dual Core 64 GB VMWare ESX 3.02 VMHost #2
RHEL4 4 x AMD 2.8 GHz Dual Core 64 GB VMWare ESX 3.02 VMHost #3


Virtual Machines: hosted on a 3-way ESX cluster (VMHost #1-3 above)
OS CPU RAM Disk Services
RHEL4 1 VM CPU 2 GB 20 GB VM HD Index Service, syncGTS
RHEL4 1 VM CPU 1 GB 20 GB VM HD GME, caDSR, EVS, syncGTS
RHEL4 1 VM CPU 1 GB 20 GB VM HD GTS Master, syncGTS
RHEL4 1 VM CPU 1 GB 20 GB VM HD GTS Slave, syncGTS
RHEL4 1 VM CPU 4 GB 20 GB VM HD caGrid Browser, caGrid Portal Web, syncGTS
RHEL4 1 VM CPU .75 GB 20 GB VM HD FQP, Workflow, syncGTS
RHEL4 1 VM CPU .75 GB 20 GB VM HD Grid Grouper, syncGTS


  • A note about virtual machines: Several services require synchronization system clocks (due to time-based operations, such as the Index Service, or time-based certificate constraints, such as Dorian or the Authentication Service). In some deployments of Virtual Machines, time drift has been observed under a moderate load. For this reason, Dorian and Authentication Service were not hosted on virtual machines in the production environment, though we have successfully deployed them on Virtual Machines in other deployments.
Personal tools