10.4. Globus grid services

The following sections describe a few of the services that will be available with Globus Toolkit V3. Again, this is early information, and may change before general availability. It is presented here to provide the reader with a flavor of what the services in Globus Toolkit V3 will look like.

10.4.1. Index Services

Index Services is like the MDS feature in GT2. It provides information about the grid services in XML format. Unlike GT2, there is no need for a GRIS because each service has a set of information associated with itself. This information is stored in a standard way, making it easy to retrieve and understand the service data. Each service is required to report common service data and any additional data is optional, allowing users to get a standard set of information from any grid service.

10.4.2. Service data browser

This allows users to view the details of the grid services available and the Web Server Description Language of those services, within a GUI.

10.4.3. GRAM

With the Globus Toolkit V3, users will submit jobs by the way of Web services. The GRAM architecture will be rendered by OGSA via five services:

• The Master Managed Job Factory Service (MMJFS) that is responsible for exposing the virtual GRAM service to the outside world. The Master uses the Service Data Aggregator to collect and populate local Service Data Elements, which represent local scheduler data (freenodes, totalnodes) and general host information (host, cpu type, host OS).

• The Managed Job Factory Service (MJFS) that is responsible for starting a new MJS instance. It exposes only a single Service Data Element, which is an array of Grid Services Handles of all active MJS instances.

• The Managed Job Service (MJS) that is an OGSA service that can submit a job to a local scheduler, monitor its status, and send notifications. The MJS starts two File Streaming Factory Services for stdout and the stderr of the job. Their GSHs are stored in the MFS Service Data Element.

  A Service Data Element (SDE) is an XML element containing information about a service that is identified by name and type, which may contain any XML information and that may be queried or subscribed.

• The File Stream Factory Service that is responsible for creating new instances of a File Stream Service.

• The File Stream Service that is an OGSA service that given a destination URL will a stream from the local file the factory was created to stream (stdout or stderr) to the destination URL.

The user then signs this request with her GSI proxy credentials and sends the signed request to the Master Managed Job Factory Service (MMJFS) on the resource that provides a function similar to the Globus Toolkit 2 gatekeeper. The MMJFS still determines the local account in which the job should be run by using a grid-mapfile, as for the Globus Toolkit 2.

Grid Resource Identity Mapper (GRIM) is a setuid program that accesses the local host credentials and from them generates a proxy for the Local Managed Job Factory Service (LMJFS). This proxy credential has embedded in it the user’s grid identity, local account name, and local policy about the user.

LMJFS invokes a Managed Job Service (MJS) with the request and returns the address of the MJS to the user. The user then connects to the MJS.

The Globus Toolkit V3 RSL language will be described in XML format even if the functionality remains similar to Globus Toolkit 2. The Managed Job Service will translate it into scheduler-specific language.

Example 10-1. Globus Toolkit V3 RSL example

<rsl: rsl <!--- insert GRAM RSL Namespace ---><gram:job>
      <gram:executable>
         <rsl:pathElement path="/bin/ls"/>
      </gram:executable>
      <gram:directory>
         <rsl:pathElement path="/tmp"/>
      </gram:directory>
      <gram:arguments>
         <gram:argument>-l</gram:argument>
         <gram:argument>-a</gram:argument>
      </gram:arguments>
  </gram:job>
</rsl:rsl>

Managed Job Factory portType

The Managed Job Factory Service defines an OGSI/WSDL interface for submitting, monitoring, and controlling a job. It is used by a GRAM client to submit a job.

The CreateService operation of the Managed Job Factory portType prepares a job for submission. It takes as input parameter RSL xml document specifying the job to be run and returns the Grid Service Reference (GSR) to MJS as an output parameter that is a WSDL definition of the MJS instance.

The Service Data Element of the Managed Job Factory portType lists the GSHs of MJS instances.

Managed Job portType

The Start operation submits a job.

The MJS will clean up everything when the job is destroyed: It cleans up directories, files, Gass cache, and so on.

The Service Data Elements include:

• Job status

• GSH to File Stream Factory Service for job’s stdout

• GSH to File Stream Factory Service for job’s stderr

File Stream Factory portType

The CreateService operation prepares to stream a job’s stdout or stderr to a destination URL. The input parameter is the destination URL and the output parameter returns the GSH.

The StartStreaming operation starts the streaming to the destination URL.

GRAM Client Interface

Both a C and Java API will be provided in the Globus Toolkit V3 to communicate with the Master Job Service.

An API translator will be also provided in the Globus Toolkit V3 for the Globus C API, the Java and Python Cog Kit, that will translate the Globus Toolkit 2 RSL format into the Globus Toolkit V3 RSL format based on XML.

10.4.4. Reliable File Transfer Service (RFT)

This is an OGSA-based service that provides interfaces for controlling and monitoring third-party file transfers using GridFTP servers. It is similar to the GT2 globus-url-copy tool.

10.4.5. Replica Location Service (RLS)

Large sites often replicate data to provide quick and easy access to data. Distributing replicas of the data reduces data access latency. RLS maintains a registry of information about where replica data resides and makes it easy to find data locations.