More Features Added to GridLink (February, 2007)
Coincident with the release of AXIS 12.0 in February, 2007, a new version of GridLink has also been released with two exciting new features: Active Backup Functionality and System Batch Support.
AXIS GridLink is a Grid solution to managing a processor farm used for distributed processing of AXIS batch calculation runs. It offers a comprehensive set of tools that help you to create and manage a more efficient and secure production environment. The AXIS GridLink Module consists of two primary components:
- AXIS GridLink Controller, which must be installed and run as either a Windows service or an active application on each Windows processor operating as part of a connected server farm, and
- GridLink Module Utility, which can be installed on any Windows desktop remote to the server farm, for the use of the Administrator who manages the farm and the operation of GridLink and AXIS on that farm.
Two new powerful features have been introduced in AXIS GridLink version 2.30 coincident with the release of AXIS 12.0.
Regular archiving of production datasets and other contents of server farms is a necessary process, yet presents challenges when the farm is constantly or unpredictably required to be available for user applications. The archival process will involve backing up datasets and other files from folders where AXIS applications under AXIS GridLink control may be performing calculations. This presents the following potential problems:
- User data that is being modified by AXIS may not be fully available to the backup utility application, This may cause some folders to be backed up in partial form, which will be unusable when restored.
- The backup software may lock various files during backup process, which will then interfere with AXIS runs attempting to access those files. This can cause various errors or even crashes of the AXIS application.
The recommended and safest way to perform backups on a GridLink processing farm is to copy AXIS data (e.g. AXIS dataset or import/export databases containing reports) while it is not being accessed to a special archival location from which it can then be safely archived to tape or to another media. Previously the administrator would have to schedule this copy process while the farm was inactive, or temporarily suspend farm processing.
GridLink version 2.30 now has a built in backup functionality that allows the administrator of the processor farm to schedule and automatically launch jobs to copy or zip selected folders from working GridLink partitions to a specified location under GridLink control.
GridLink uses real-time information from its job queue manager to avoid collisions between the backup processes and ongoing AXIS calculations while this scheduled backup process continues. Active AXIS jobs will still continue once the scheduled backup commences, but GridLink will temporarily suspend launching any new jobs in the queue until the backup procedures are done. Meanwhile folders used by running jobs will be skipped during the backup process; however, GridLink will then attempt to backup these datasets at the end of the backup procedure if there is sufficient time allocated.
Multiple backup jobs can be scheduled and a specific window of time can be defined for each backup job to make sure that the process is not overlapping with externally scheduled archival processes such as transferring the copied or zipped files to a tape backup. If the AXIS backup finishes within the scheduled window, the GridLink controller reactivates the queue immediately. If the backup is not finished within the window, the backup job is terminated to avoid conflicts with the external archival process, and records the early termination in the system log.
A full report of the last backup job execution will be stored for the farm administrator to review.
The Active Backup feature is controlled using the GridLink Module Utility, through the Backup Manager option on the Tools menu.
Note that while Active Backup in available in version 2.30 of GridLink, it does not require AXIS 12.0 to be installed. For instructions on how to use it please refer to the AXIS help topic:
AXIS GridLink Active Backup Manager
System Batch Execution on GridLink
Distributed processing focuses on sharing various batch recalculation and projection tasks over all available processors on the server farm. Dataset Batches which may have included such distributable tasks have been supported in GridLink for some time, but System Batches, which can operate on datasets and perform other higher level tasks such as backup and restore operations, could not be executed within GridLink.
AXIS GridLink version 2.30 in combination with AXIS 12.0 now supports the execution of all system level batch jobs, including:
Backup / Restore
This enhancement enables you to configure and run entire production sequences including multiple datasets and calls to external programs and procedures on the GridLink processor farm without user interaction.
The most powerful of the above System Batches is the System Formula that allows the user to define production procedure code in AXIS Script language. System Formulas can include such steps as restoring the datasets, manipulating the source files, performing the runs, analysing errors and backing up the datasets and results after the runs. Various steps can be conditionally executed according to the status of previous steps or other circumstances at the time of execution.
To submit system level batch jobs to a GridLink processor farm, you may use the same simple method as for Dataset Batch jobs; that is, just select the "Run As Remote Job " option from the Action Menu while viewing the list of System Batches in the Dataset on your desktop. Alternatively, a System Batch may be submitted to the farm using a command line approach.
For more information on using System Batch Execution in GridLink, or on GridLink in general, please refer to the AXIS help text, the related articles listed below, or call GGY for assistance.