Clients ask us this question frequently: “…Should we stick with our business applications on SAS 9.4, or should we take the plunge into SAS Viya? I’ve heard there are some functional differences between the two, and some of my colleagues are concerned that Viya implementations still feels new and a little risky. As a professional designer and implementer, what are Corios’ assessment of the risks and challenges (and benefits)?”
Everyone’s mileage will vary, but here is our experience with migrating three clients from traditional SAS to Viya on AWS. Our clients include a credit card issuing-bank, a commercial bank, and a business financial data clearinghouse.
The good news is, migrating the SAS business application is not the challenging part. Yes, there are differences between how SAS 9.4 and SAS Viya work, but moving the business application from one major SAS release to the other was relatively straightforward. The real challenges emerge on the security and integration side of things. This is because clients want to pursue cloud migration and system integration for reasons that include data and analytics but are broader than the SAS application itself. In particular, we see the thorny issues with privacy, networking, authentication, tenancy and the scalability of data movement.
As specialists in machine learning and analytics modernization, we can’t rely on those analytic skills alone to address these challenges. We’ve benefitted greatly from our investments in systems integration, cloud architectures, data migration, information security and industry standards compliance. The combination of all those skills helped us guide our clients on the tradeoffs of cost, complexity and security and overcome these challenges.
Credit portfolio optimization: moving the business application to Viya
For one of our credit card issuer clients, back in 2012-13, we built a business application on SAS 9.4 to help them optimize their credit card portfolio risk and profitability. This optimization engine uses mathematical simulation and scenario optimization of various strategies: pricing, product features, customer segmentation and credit risk policy. The original release of this portfolio optimization application used SAS 9.4 for the big data functionality, and JMP for the in-memory simulation and optimization functionality, plus a very visually-intensive user interface.
Over the next 7 years the client continued to use this as their primary system for customer acquisition strategy and optimization. When they decided to upgrade their systems to SAS Viya, they asked us to upgrade this business application. We updated the entire application to consolidate all the functions inside SAS Viya, retired the JMP components, and moved the interactive user interface into a combination of HTML5, Java and JavaScript on the front end and the SAS Viya Job Execution Server (JES) on the back end. Our development team found Viya was relatively easy to integrate with the user interface development framework we designed, and was more flexible than the SAS 9.4 Stored Process servers and APIs that we have used in other client engagements.
Overall the engagement required about 3 months to complete, and the client is off and running with the new release of their application.
Commercial bank next best offer application: moving the business application to Viya on AWS
One of our commercial banking clients hired us to build a next best offer (NBO) application in 2019, which we constructed in a private data center using SAS 9.4. About a year later, Corios had decided to migrate all our secure applications for banking clients to AWS. This is because we could get the same or better security, reliability and privacy capabilities at a substantially lower cost for our own internal infrastructure and for our clients, and benefit from AWS’ cloud service development roadmap. In 2020 we started migrating the NBO application to AWS, including the data storage tier (Microsoft SQL Server) and the ETL and analytic model factory (SAS).
Migrating the SAS components and business application was straightforward; where we encountered more challenges was on the infrastructure side of things. AWS provides a Direct Connect service to provided a dedicated network connection for reliability, which we leveraged. Due to different configurations on the bank’s on-premise network, and the Corios managed service on the AWS side, it took some iterations to coordinate with the network professionals on both sides of the handshake to design, test and validate the networking protocols. We also found that database migration was slightly more complicated than in typical database migrations, since we selected the Amazon RDS service as a managed database server, which makes many operations simpler but introduced a little additional complexity for the data schema migration.
One item of note is that the environment used for this application is in the scope of Corios’ SOC2 and PCI-DSS audit, so the options available for migration and data ingress/egress are more limited than in many conventional projects. All told, the migration went fairly easily once we’d straightened out the network interconnects and security protocols.
Business financial data clearinghouse: Migration to SAS Viya on AWS
This is probably the most technically challenging project of the three described in this article, because it has the most advanced capabilities and requirements. Our client is a clearinghouse for business financial data, used by banks, insurance companies, investors and hedge funds, so naturally security and reliability are prominent (as is with most of our clients). Picture a “Bloomberg Terminal” for private and public businesses, where the business data is searchable, indexable, and predictive and you’ll get a pretty good idea about this business application.
The security, authorization and reliability aspects of this application are of primary importance to Corios and our client. Some of our client’s customers will use a multi-tenant environment where all common data is shared by the clearinghouse but is logically partitioned. Other customers can choose a single-tenant environment where they can bring their own data and overlay it on our client’s data.
One of Corios’ challenges is to support some very advanced interactive capabilities, which don’t come out of the box with SAS Visual Analytics (“SAS VA”). Our experience with augmenting SAS VA has proven that we can confidently extend VA via JavaScript, API integration and open source extensions. SAS VA provides a ton of functionality out of the box, but many clients will push the envelope beyond what is possible without extensions.
Furthermore, Corios needs to ensure all operations align with the SOC2 and PCI-level data security standards that we provide to our financial services clientele. On the authentication and authorization side of things, this includes providing multifactor authentication and record level data security.
As mentioned above in the first two case studies, implementing SAS Viya on AWS is not the hard part, as the SAS elements are well understood and we’re staying in the boundaries of what is reliable. The challenges are supporting all the other capabilities, including the surrounding security, tenancy, reliability, user experience and e-commerce capabilities that our clients need.
How are you modernizing SAS?
I’d love to hear from you about your own SAS modernization goals and objectives. You can read more about our capabilities under the Corios Rosetta banner, and you can also reach out to me directly at president@coriosgroup.com. Contact me today!
