With files from Lynn Greiner.
On Sept. 27, SAS officially brought its Viya analytics platform to the Microsoft Azure Marketplace. The service enables Azure customers to use the solution on a pay-as-you-go basis with the click of a button.
Bryan Harris, SAS chief technology officer, explained that this marks a significant move towards helping organizations move to the cloud and adopt AI. Harris said that while SAS’ digitally born competitors started with the cloud, SAS’ long history means that many of its customers were running their solutions on-premises, and that preparing for their transition to the cloud was no easy task.
“We’ve had to do the inverse of what many of our competitors did,” said Harris. “Because we have such a large [on-prem] customer base, we want to make sure that those customers have a path of least resistance to accelerate the adoption of AI and move to the cloud.”
To adapt, SAS had to put in some elbow grease to prepare its solutions to not only be easily enabled, but also offer the same or better performance and reliability in a cloud-native environment.
Related: AI and wine are not that different, says SAS’s director of AI ethics
Responding to the dynamics, SAS had to revamp its solution’s security, performance, and operating environment to best suit these new criteria.
Harris noted that new concerns arise when organizations move to the cloud. For example, when software is running in containers in a server cluster, it may need to limit the use of elevated access privileges because there may be other software running inside the same cluster. This, in turn, restricts some of the functions users are familiar with in a bare metal environment. And that’s just one of the hurdles software companies are facing in their transitioning process.
Performance is of course another key concern. “Everyone just thinks that if you go to the cloud it gets faster and cheaper, that is not true,” Harris said. “In fact, if you don’t do it right, the cloud becomes more expensive. This is why the lift and shift model never works, because people take these assumptions…move it over to the cloud, and they get upset about the bill they get.”
Organizations are still grappling with these challenges. Harris cited a recent IDC study that revealed only 40 per cent of software companies have successfully migrated their operations onto the cloud.
To best prepare its solutions for the cloud, SAS had to refine the way the software operates. It has to be able to quickly turn on, shut down, and scale for easy management and to reduce costs.
And the conversation around cost is still a difficult one. Companies are still having trouble balancing performance and cost.
SAS had to consider all three factors when focusing on the cloud. Harris noted that SAS solutions use an architecture that works with all popular clouds like AWS, Azure, Google, and Red Hat OpenShift.
This announcement signals a deepening of the ongoing partnership between SAS and Microsoft that began in 2020. SAS, the world’s largest private software company, has been aggressively pursuing the cloud and AI. Similarly, Microsoft has been equally ambitious in its cloud expansion efforts. SAS Viya integrates with Azure services through the entire analytics stack to achieve better data discovery, modelling, and execution of analytics models.
Alice McClure, director of product marketing at SAS, described the integration, and the partnership with Microsoft overall, as a win-win scenario. In addition to benefits for existing Azure subscribers, it also entices SAS users looking to sign up for the cloud service.
SAS has positioned Viya on Microsoft Azure Marketplace as a pay-as-you-go offering at a starting rate of US$0.55 per hour, based on the utilization of the virtual CPU clusters.
“Taking those capabilities and making them immediately available in this way, having an app resource available to help you onboard and get started…is something we’re very proud to have provided within this offering,” said McClure.
Delving deeper into the business model, McClure said customers can sign up for the service whenever they’re ready. When they do, the service is provisioned via an automated deployment in their Azure tenant. The cloud provider then meters usage and bills the customer. SAS has released learning resources to help users efficiently use the new service.
Looking toward the future, Harris promised that SAS has big plans to help its customers, especially for those facing variable and consumption-based costs.
“We’ve got some things coming that will be very, very groundbreaking on the performance side, and around reducing cloud costs,” he said. “There are bigger ones coming in 2023 where we’re gonna see things in the order of 5 times, 10 times of returns, minimizing the need for customers to use GPUs, which are expensive, and using traditional CPUs for the same answers and the same performances.”
