Data be nimble,
Data be slick,
If data dumps can access the cloud real quick.
It’s a terrible nursery rhyme, but one that Jim Goodnight, the co-founder and CEO of SAS Institute might appreciate after showcasing the analytics software firm’s new cloud platform, Viya, at its Analytics Experience conference this week in Vegas.
With the introduction of Viya, the Cary, N.C.-based SAS hopes to appeal to a new audience and enable its existing customers with large datasets to get answers faster. While SAS announced its Viya platform for the first time in March, it was at this week’s conference that it unveiled the first applications to be made available on the platform, demonstrating the target audiences that it sees as key analytics users and giving many existing users motivation to reach into the cloud, if they haven’t already done so.
Cognitive computing capabilities are also at the heart of the new products showcased at Analytics Experience. While SAS has been working in this space since the ’90s, it can now apply more powerful deep learning algorithms by leveraging cloud architecture. That will also lead to more natural ways for users of the software to interact with it, meaning that in the future you could ask SAS software a question spoken out loud – just as you might do with Siri today – and get an answer pulled out of your deepest data troves within seconds.
“Cognitive is just wonderfully challenging and a state of the art place to be right now,” Goodnight says. “We enjoy that challenge of new things. We’ve been doing lots of little pieces of it for some time and now we’re pulling it all together.”
SAS is working with neural net architecture that passes data from one layer of code to the next, mimicking the process of neurons in the human brain, to try and achieve better object recognition and capabilities like text-to-voice. While there’s no SAS software that you can chat with over a coffee just yet, the Visual Investigator software shows how the analytics firm is appealing to users not comfortable coding.
Analytics for analysts
“There’s a lot of people that do not like to program or learn the necessary syntax,” Goodnight says. “By providing the visual interface we make it much easier to explore.”
The tool can help analysts discover hidden relationships be exploring networks of interconnected data, all represented with hub-and-spoke graphs that are easy to interpret. A point-and-click interface enables analysts to set up event-based triggers and alerts, to prioritize the data that the organization is dealing with.
Here, SAS is clearly attending to the needs of the business user, says Ray Wang, principal analyst at Constellation Research.
“You want to be able to do more with data – incorporate it into a forecast, model it. How do I consume this stuff?” he says.
Analytics for data slingers
Then there’s the Visual Data Mining and Machine Learning tool also available on Viya that is squarely made for data scientists. A coder’s dream, it supports open languages Python, Java, and Lua, and supports interface with REST APIs too. Beyond the open approach, there’s good reason that many users already programming with SAS 9.4 on a local installation may want to consider making Viya a part of their analytics diet.
“We’ve got a huge install base on Hadoop and we want to encourage IT to not just use this as a storage area, but as a computing and analytical environment, Goodnight says. “Viya allows you to do this.”
Apache Hadoop is an open-source software framework used for distributed storage and compute with large data sets. While it allows for all the compute power available on-premises to be clustered together and pointed at corporate data sets, opening up those datasets to cloud compute completely changes the scale of operations.
“What they can’t do on-premises is access compute power for deep learning,” Wang says of SAS 9.4 customers. “These data sets keep getting bigger and the algorithms are becoming more complicated.”
A customer with a dataset of just 1 TB in size could benefit from turning on the Viya taps, Goodnight says. By cutting out all the network traffic, analytics users avoid latency and can work with thier models in real time.
Besides, he adds, SAS will be able to update Viya more frequently with its latest capabilities compared to the local installations maintained by most companies. Often, corporations can require months of testing before releasing new software into the operational environment for users.
“That can be very frustrating for the users,” Goodnight says. “Just to have to wait for up to a year before they get their hands on it.”
Viya can be deployed in whatever form a customer wants, Goodnight says. SAS can provide the infrastructure required to support it, or it can be deployed to Amazon Web Services or Cloud Foundry. Even running it on bare metal is an option.
Expect to see Visual Investigator and Visual Data Mining and Machine Learning available within a month or so. Canadian pricing isn’t available yet.
