Doug Henschen on Analytics, Big Data & Smart Apps
IBM serves up data-analytic cloud services, an Apache Spark service on BlueMix, and new data-discovery capabilities within IBM Watson Analytics. As for that Cognos update? It’s a half step to self-service BI.
IBM’s announcements its IBM Insight 2015 event this week in Las Vegas weren’t all about cloud, but those were the ones I found to be the most interesting during an opening-day keynote otherwise focused on highlighting recent accomplishments and “Insight economy” vision statements.
The three cloud announcements that caught my attention concerned:
Here’s a quick rundown on what I liked and why.
Insight Cloud Services
IBM announced a set of data and analysis services aimed at bringing contextual insight to next-generation web and mobile apps. The data is from partners Twitter and The Weather Company and more than 150 other sources including public sources such as the U.S. Census and Bureau of Labor Statistics. [Note: The day after this post, “IBM announced that it has entered into a definitive agreement to acquire The Weather Company’s B2B, mobile and cloud-based web properties, including WSI, weather.com, Weather Underground and The Weather Company brand. The Weather Channel (TV business) will not be acquired by IBM, but will license weather forecast data and analytics from IBM.] The deep analytics draw on APIs borrowed from IBM Watson and include data-analysis and language-processing capabilities such as entity extraction, which is used to spots people, places, things and events within textual data.
The portfolio will include some 20 services, and IBM has licensed a high-scale, high-speed data-delivery platform from The Weather Channel, which delivers mission-critical weather data to airlines, insurance companies, media outlets and many other industries. Running on Softlayer, the platform is described as robust and scalable with high availability.
As for the purpose of these services, the idea is to extract, integrate and markup data and then provide contextual analysis that brings meaning. Joel Cawley, the general manager of Insight Cloud Services, used the example of the simple weather data point of a 50-degree temperature reading. That’s not so remarkable unless you add the context that it’s 50 degrees in Boston in February, where it has been below freezing for the last three weeks straight. Or maybe it’s Miami in August, and temperatures haven’t dropped below 75 since May.
The emphasis with Insight Cloud Services is on delivering actionable information, according to Cawley. Weather data services, for example, could be used by utilities to forecast demand and predict service outages, by local governments to develop emergency plans, or by retailers to optimize inventories and increase sales.
IBM introduced industry-specific Insight Cloud Services last May, including IBM Demand Insights, used by retailers and others to understand the correlations between sales of specific products with weather, events, news, trends and social commentary. Customers Urban Outfitters and Costco presented here in Vegas. The IBM Market Insights service is used by consumer products and media companies to better understand customers based on likes and interests expressed in social media. This provides customer-segmentation data that can be used to improve targeted marketing efforts.
IBM also introduced a cloud-based Fan Insights Service for sports and entertainment firms. The Ottawa Senators hockey team is presenting here on how it plans to use the service. The service is aimed at predicting ticket sales and concession and staffing needs throughout the season as well as effective marketing strategies based on fan sentiment, behavioral trends and individual fan preferences.
MyPOV: These services sound compelling, though I’m a little sketchy on just how contextual data services translate into actionable insights. The temperature example (50 degrees in Boston versus Miami) makes sense and who doesn’t want actionable insight? But is this support-heavy approach — much like IBM’s joint mobile iOS apps with Apple – whereby supporting implementation services will be required to deliver actionable insights within applications? I’m hoping these services will be straightforward and easy to use for Web and mobile developers.
IBM Analytics on Apache Spark
Spark is, of course, the white-hot open source in-memory analytics framework that IBM promised it would back in a big way earlier this year. This managed Spark service on IBM BlueMix has been in the works for a few months. On Monday I sat in on a Spark panel that included early customers Climformatics and consulting firm SmarterData.
The service is said to include Spark Core as well as its SQL, Graph, R and MLLib machine-learning components. There’s also a Notebook user interface for accessing, loading and visualizing data with drag-and-drop functionality. The service accesses data from BlueMix cloud services including Cloudant, DashDB, Streams and the DataWorks data-transformation and cleansing service.
MyPOV: I’m a big fan of Spark for its in-memory performance and analytic versatility, and this new service is early proof that IBM is following through on its commitment to develop Spark for the enterprise. IBM’s new service gives developers a much-needed option (besides Databricks) for learning and deploying Spark-based applications in the cloud. It runs stand-alone on cloud storage in IBM’s cloud, with no requirement to also run Hadoop (which should make things easier). It’s a pay-as-you-go, big data service that’s ripe for the times.
IBM Watson Analytics and Cognos
IBM Watson Analytics is IBM’s intuitive, cloud-based app with natural-language question-and-answer capabilities and smart, automated recommendation for visualization and analysis. Cognos is, well, the aging business intelligence suite born in an earlier era, but IBM has announced a significant facelift.
A lot of the sexy stuff inside IBM Watson Analytics was actually developed by the Cognos and SPSS teams, but IBM decided to add a dollop of Watson and serve it up as an independent, cloud-based product. Watson Analytics even has self-service predictive analytics capabilities, which I detailed in this report.
The “what’s new” in this IBM Watson Analytics refresh includes a new Expert Storybooks feature for data discovery. Developed in collaboration with nearly a dozen partners, Storybooks help users spot the most relevant facts, patterns and exceptions in data. A Deloitte-developed Storybook, for example, measures the effectiveness of incentive programs, while one from The Weather Company helps users understand how weather impacts revenue trends. IBM also introduced a Secure Gateway for Watson Analytics for accessing on-premises data. And it added connectors for DB2, Informix, Netezza, IBM SQL Database, IBM dashDB and a variety of popular third-party data sources.
As for the latest upgrade of Cognos, IBM introduced an extensive user-interface refresh aimed at consolidating overlapping functionality and bringing self-service capabilities to report and dashboard consumers and creators. It also introduced “Intent-Driven Modeling” that interprets what you are after based on search terms. Unfortunately, the deeper you go, the more you see all the complexity of the underlying product. And IBM has done little to streamline administration and the heavier aspects of data-management.
MyPOV: IBM calls Watson Analytics its tool for citizen data scientists (going after Tableau) while Cognos Analytics is, well, the legacy product. I can see where existing Cognos customers will appreciate all the self-service improvements in this upgrade, so maybe it will stem the rate of attrition. But I wouldn’t expect a flood of new customers.
Watson Analytics looks like the future for IBM, but it’s up against Tableau, Qlik, and a host of new and revamped cloud options from the likes of Amazon, Microsoft, Oracle and Salesforce. It’s notable that Amazon is addressing the complexity of data analysis starting with the back-end data layer with its recently announced Quick Sight service. That service is obviously far from proven, but it just may be that simplicity and ease of use has to start with core data management.