GridLink - New Batch Type "Grid Job Set" (February, 2017)
A new AXIS batch type "Grid Job Set" has been introduced. This batch type is designed as a solution to address scalability limits.
In case your dataset does not contain enough distributable tasks to use the entire GridLink farm, you can use a Grid Job Set batch to simultaneously run multiple independent batches from the same dataset. E.g. you have multiple subfund batch testing batches representing different business lines within the company. No individual batch has any dependency on another. In the past, users had been cloning the datasets multiple times so they could run all of them on GridLink servers at the same time. The dataset cloning and consolidation of results are not necessary anymore with this new batch type.
Note that by design only import / export databases will be transferred back to the dataset, so users will not have access to any of the projection / block / daily hedge results within the dataset. Also, each batch must output to separate database files.
Batches inside this new batch type will all be submitted to a multi-queue GridLink farm and the user can monitor progress over multiple queues.
Each sub batch within the Grid Job Set will be executed as usual. Distribution by batch during the batch run time is supported.
Note that such a batch set can run only on a (multi-queue) GridLink farm, that is: a Standard Distribution model will not be applicable to this type of batch set.
This feature requires GridLink version 7.60. The user will get an error at submission time if the batch is submitted to a GridLink controller older than 7.60.
Further details on how to use this new batch type is available in the help text:
GridLink Version: 7.60