In our world of ever-increasing data volumes and desires to interrogate that data in more computationally intensive ways, our methods to perform these operations has had to evolve rapidly.
We just don’t have the compute power in individual instances easily available to conduct the level of analysis we’d like. The days of large, expensive, standalone, program-specific computing resources have gone, and focus has shifted towards language-agnostic grid computing and cloud-based architectures.
As a leading provider of analytics tooling, SAS has created several products to service this need, and allow their customers to extract the maximum value possible from their data. These offerings have evolved from a third-party grid manager, which runs on expensive on-premises dedicated hardware, through to the latest incarnation (SAS Viya v4) utilising Kubernetes containers, to provide cloud native, workload-agnostic architectures, allowing highly scalable, highly available systems to become the norm.
SAS Grid – Past, Present & Future
Initial grid offerings utilised Platform LSF, a respected workload management tool supplied by IBM. This was later joined by the possibility of utilizing YARN for workload management within Hadoop, ideal for larger applications, to allow workload balancing across a range of ETL and analytics products, extracting much greater value from the commodity hardware used to host Hadoop, and one of the first occurrences of SAS embracing the use of other analytics and data management products.
This was followed by a SAS implementation of a grid manager, written in house, and supplying some support for other products, and beginning to embrace the world of open source, a growing source of software across most industries. This SAS Grid allows batch processing of any jobs, but interactive support is limited.
The very latest grid-like product from SAS utilises Kubernetes for load-balancing, scalability and resilience - as is used by many of the world’s largest organisations for a very wide variety of applications. Kubernetes (“k8s”) provides a highly available, scalable, and resilient technology landscape with low running costs and great use of computational resources.
The current solutions offer two great solutions to those utilising either on-premises or cloud-based computing, but for the future it’s tough to look past Kubernetes. The flexibility and resource utilization is hard to beat, although the latest thinking of serverless technology is potentially a game changer….
If your organization is interested in exploring how you might utilise cloud, Kubernetes and containerization in your statistical computing environment, check out d-wise’s Solution Build offering.
