How is GlaaS different?
GlaaS vs. Traditional GridLink
Traditional GridLink farms are designed around fixed size grids with one or more queues. In traditional GridLink farms, jobs submitted to the same queue run in sequence while jobs from different queues in the same farm can run in parallel, effectively competing for processing resources. GridLink allocates capacity by splitting the fixed number of processor cores found in the farm based on job priorities.
GlaaS uses the massive processing capacity of the cloud to overcome these constraints. The GlaaS engine automatically provisions a separate GridLink farm in the cloud for each submitted job based on the size limits requested by the submitter. This provides significant benefits:
- There is no need to wait for other jobs to complete; the newly created dedicated farm is ready in minutes and your job starts running with as many cores as you have decided to allocate
- Any number of jobs can be submitted and run in parallel since each job will have its own farm
- There is no competition for network, CPU or disk bandwidth between jobs as there is when jobs are running on the same traditional farm
- Each farm can always be made the right size for the job; total usage charges for a job with efficient distribution will be no greater for a large farm than a smaller one, but time to completion can be drastically reduced. You will pay only for the core-hours of allocated compute capacity used (there are no fees for storage or data transfer involved with the GlaaS service).
As a result, when you use GlaaS, you will have more predictable run times for production jobs which must meet your reporting deadlines, and your incremental costs are minimized. In addition, unexpected demands for additional runs can be easily met without disrupting other production priorities.
There are several other advantages of using GlaaS compared to traditional alternatives:
- GlaaS uses the latest cloud networking, storage technology and virtual server instances which addresses the challenges of relying on aging infrastructure
- You may conveniently access GlaaS from any authorized computer in your on-premise, cloud based or hosted AXIS environment; GlaaS integrates with your current AXIS EnterpriseLink installation for interactive and automated job submissions and monitoring
- GlaaS server farms are deployed in multiple cloud regions ensuring redundancy while also respecting data residency requirements
- The GlaaS service is fully managed by the Moody's Analytics Cloud Operations Team
GlaaS vs. GridLink Bursting in the cloud
Some clients have already arranged for GridLink to be used in a cloud-based server farm that they provision themselves with their chosen public cloud provider. This has been helpful for those large or urgent jobs where more traditional server farms, on premise or hosted, have not provided sufficient capacity. This is typically referred to as GridLink bursting in the cloud, and GridLink may be licensed to manage these cloud-based GridLink bursting farms on a core-hour basis. These farms are typically setup with a defined number of cores for the duration of the farm life time. The end user is responsible for turning the farm ON and OFF and AXIS jobs are submitted to corresponding GridLink queues similar to the use of traditional on-premise GridLink farms.
Our new GridLink as a Service offering is different:
- GlaaS provides a fully automated way to run jobs in dedicated cloud based farms that are provisioned for each individual job, thus greatly improving the efficiency and performance while reducing the overall cost of using the cloud for AXIS runs
- GlaaS offers all the important advantages over bursting that are described above in the comparison to traditional GridLink farms
- Switching to GlaaS eliminates the need for you to negotiate, setup, maintain and pay for infrastructure in the public cloud that is needed only for AXIS runs
- GlaaS fees paid to Moody's Analytics are "all inclusive" while our bursting fees typically cover only the GridLink software license charges, leaving all other infrastructure arrangements to you to manage and pay for