According to COO and CTO Oliver Schabenberger, 2019 was a transformative year for SAS. The company is responding to the same tectonic shifts impacting the entire enterprise software provider ecosystem, but with some SAS unique twists.
For instance, like many of its counterparts, SAS is making the shift from traditional licensing to subscription. But unlike most enterprise software firms that sold perpetual licenses, SAS's previous license already had similarities to the subscription model in that clients rented rather than purchased the software. The company, whose growth flattened to about 1 – 2% annually over the previous five years, didn't report revenues this year. That's akin to counterparts that went private, except that SAS was already private. SAS is also biting the bullet and rewriting its next-generation software stack into a cloud-native architecture, while counterparts such as Informatica, have already gotten there. And the company is fighting for mindshare with the next generation of practitioners in a market that, as we noted last year, is now crowded with trendy upstarts.
Let's start by comparing some benchmarks from last year. A year ago, we reported that Viya sales began its first sharp upward climb as sales doubled. That reflected the filling out of the Viya portfolio with data preparation, econometrics, model lifecycle management, optimization, data mining and machine learning, text analytics, computer vision and natural language interaction. This year, Viya grew 47% and added features such as single-click model deployment, more natural language interaction with Visual Data Mining, along with more AI and machine learning capabilities. We reported last year that SAS's cloud business jumped 30% (largely attributable to Results as a Service), but the company didn't report results this year.
Putting skin into open source
As we previously noted, Viya provided the path for SAS to working with open source software. The guiding notion is, use the algorithms and languages (read: Python) of choice, but leave the lifecycle management and mathematics to SAS. We've previously ranted that SAS should contribute to open source; later this spring they plan to put their money where our mouth has been, announcing their first contributions.
SAS is taking yet another step on the lifecycle end, as the next release of Open Model Manager that has been included as part of Viya (shipped in February) will also be available as a standalone, cloud-based tool that won't require SAS software. Clearly, SAS is hardly the only provider offering lifecycle management of models, but as enterprises increasingly rely on models to help them make decisions for which they will be accountable to regulatory authorities, this will become a more critical functional gap for organizations to fill. SAS's appeal is that it already has the grounding with enterprise-scale management models. And as SAS exposes more model execution services in the cloud, standalone tools like Open Model Manager could become gateway drugs providing access to those services.
The Viya transition
SAS is clearly staking its future on Viya, and is directing all new feature development there. So, with the classic SAS suite at the 9.4 version, we do not expect that there will be a 10.0. Cutting to the chase, the upcoming 4.0 version of Viya will be fully cloud-native. It will migrate from its current architecture, to become multi-cloud with a full shift to a Kubernetes-based container architecture. That will require porting from the current 3.5 generation of Viya, as the new version will be on a different architecture – the upgrade could not be made in place while the system is running.
The good news is that, while 4.0 is a major shift in architecture, at code level, Viya's existing microservices-based design should mitigate the disruption. But for classic SAS 9.x, the migration path is still a work in progress. They have defined the process, but there's more work to go on the tooling. We'll have a lot more to talk about when SAS releases Viya 4.0 in the spring.
Product direction going deep and wide
As to future product direction, SAS is charting a couple divergent paths. One is focused on industry solutions. At an analyst event last week, SAS provided deep dive glimpses into government, manufacturing, retail, and financial services. For instance, in retail, SAS's solution can gather data across many sources, reflecting the omnichannel realty of retail to generate a unified view of demand, and then provides functionality that segments customers, plots demand for specific items or categories across specific demographic segments, then translates that to sales forecasts against which the impacts of potential promotional pricing schemes can be modelled and predicted. Based on demand, SAS's retail solution can then be used to optimize the product inventory mix for each brick and mortar store location. Schabenberger sees industry solutions accounting for 40 – 50% of SAS's revenues. And, then next major phase of development for Viya will be in bringing existing industry solutions over to the new platform, and developing new ones.
SAS's other goal, embedding its software in industry solutions, is more ambitious. Schabenberger terms it "analytics hidden in plain sight." This is essentially an OEM strategy that would be akin to the "Intel inside" campaign made famous by the chipmaker. It's more ambitious, not in terms of whether it will comprise most of SAS's' revenue, but because of the partner commitment it will require. SAS has published APIs, and it scored its first major success a year ago with a Siemens partnership for embedding SAS streaming analytics and machine learning capabilities into Siemens MindSphere IoT platform.
SAS's 40+ years in business guarantees that it will be known for its legacy, and that legacy is both the source of its competitive advantage and a hurdle in winning mindshare from the next generation of data scientists.
There's no question that SAS has critical mass presence in the enterprise. Its portfolio has the depth and breadth built from years of industry engagements that have now translated into industry solutions. Its capability to operationalize analytics at complexity and scale has been proven inside the data center. As established provider, SAS built a ready constituency inside large enterprises, but as new data scientists come out of school, they have voted with the feet and budgets for open source technologies that, in their perceptions, given them more portability on their resumes. SAS is only now, not just interoperating with open source, but about to more actively engaged with it. It's about time.
As established provider, SAS has not always been the easiest company for customers to deal with. But SAS is no longer the only game in town. Like household names of enterprise software such as Oracle or SAP, it has had to change the way it engages customers. The flattening of the new subscription licensing structure and the simplifying of support layers are examples of the homework that SAS has done, and still has cut out for it.
When it comes to cloud, SAS's established customer base kept the company from being out front until now. The core of the base did not want to be the first guinea pigs to the cloud, but now that enterprises have gotten their feet wet, that base wants to be, as a SAS product exec characterized it, "fast followers." Of course, it certainly helps that there is an established operating environment – Kubernetes – that has become the de facto standard against which SAS could rearchitect Viya. In retrospect, it may have been fortunate that SAS did not jump the gun on cloud by designing Viya around Cloud Foundry.
But the transition of the sweet spot of its customer base will be the prime challenge facing SAS. Classic SAS in the data center won't go away because of data gravity and the fact that enterprises already have programs that are written on that core code. As the ERP folks are learning, generational change is a gauntlet because, if the migration requires major change, rival solutions become fair game. Now that Viya will finally be on a cloud-native architecture, SAS's biggest job will be developing a migration path makes it the logical target.