On Analytics, Data Platforms and Smart Applications
Amazon Web Services is following in competitor’s footsteps with Athena, QuickSight, Rekogntion, Polly and Lex. Head starts won’t matter in the face of Amazon’s scale.
Amazon Web Services CEO Andy Jassy introduced a bevy of new services and capabilities at the AWS re:Invent conference in Las Vegas this week. The new analytic and artificial intelligence (AI) services aren’t unique, but there’s little doubt they’ll be huge hits.
Jassy framed his announcements around the theme of giving enterprises “superpowers.” Examples included powerful new compute instances supporting superhero-like speed, new database services enabling “flight” from the high cost of commercial databases, and new IoT services enabling “shapeshifting” out to the edge of the enterprise.
I was most interested in the “X-Ray Vision” introductions, which included Athena and QuickSight analytic services and Rekognition, Polly and Lex artificial intelligence (AI) services. Here’s a recap along with my take on each announcement.
Athena: AWS has Redshift for high-scale structured-data analysis and EMR (Elastic MapReduce) for high-scale unstructured data analysis. The company has added a third leg to this data-analysis stool with Athena, which offers interactive SQL analysis of data in Amazon S3 (Simple Storage Service). Athena promises a simpler and less costly alternative to EMR for analyzing semi-structured data such as clickstreams, logfiles and other sparse and variable data types that aren’t easily loaded into database services. Query times are said to be in the sub-second range, even at high scale.
MyPOV on Athena: This is likely to be a handy and cost-effective option, but I’m guessing popularity and use-case diversity will depend on the depth, diversity and applicability of SQL. Just what kind of queries will it support against what types of data? Microsoft just announced a similar capability with Azure Data Lake Analytics with U-SQL querying. But the Azure Data Lake service was also just announced whereas S3 is the well-established kingpin of low-cost cloud storage. In short, Amazon has a readymade market of would-be Athena customers.
QuickSight: Amazon quietly announced the general availability of this business intelligence service a couple of weeks ago, which was surprising given the fanfare around the QuickSight announcement at last year’s re:Invent conference. In fact, Amazon has been quiet about QuickSight all year, and I heard rumblings that it’s because the project hit a few roadblocks. There’s a reason BI has been hard for vendors to master all these years, so I’m not surprised there were challenges creating a tool that “makes BI easy for all employees, regardless of their technical skill,” as AWS promised. If the data you want to analyze is structured and already available in an Amazon source (RDS, Aurora, Redshift, or a comma-delimited file on S3), QuickSight can infer data relationships and get you to visualizations in minutes. What’s more, the query speed of the built-in Spice columnar, in-memory engine is plus.
MyPOV on QuickSight: On the back end, QuickSight offers less automagical understanding of data structure than I expected, based on last year’s announcement. To get data in and integrated, there are data-prep filters, various connectors and a table-join UI that are best described as tools for power users. It’s like working with a basic, self-service data-prep tool, and will hardly be automatic. As for tapping unstructured sources, such as EMR, that was promised last year but omitted from the launch press release, so I’m guessing it’s still in the works. On front-end analysis, Amazon was originally going to use technology from ZoomData, but that was dropped somewhere along the way. When I last saw a QuickSight beta demo in August, executives talked about adding common visualization types including stacked bar, area and bubble charts – evidence of starting from scratch.
In short, I believe QuickSight remains a work in progress with room for improvement. Nonetheless, given that there’s a free tier of the service available that can be used with up to 1GB of data — the Standard Edition costs $9 per user, per month for analyzing up to 10 GBs of data — I have no doubt that QuickSight will see lots of use and that customers will drive improvements over time. That’s pretty much the way Microsoft Power BI has evolved. I’d note that none of the freemium services (adding IBM Watson Analytics to Power BI and QuickSight) have obviated the need for more powerful and capable BI options.
Rekognition, Polly and Lex AI Services: There have been many AI-related announcements this year, and AWS CEO Jassy took pains to remind re:Invent attendees that Amazon has been hard at work on AI since that advent of the company’s well known retail recommendation engine. To bring more internal Amazon capabilities to cloud developers, AWS introduced three new services. Amazon Rekognition is an image-recognition service that can spot objects (car, pencil, cat), scenes (outdoors, mountain, forest) or faces (man, woman, boy, girl) within images. It can also detect sentiment (smiling, frowning, angry) and, with training, facial recognition (this is Jeff Bezos, that is Satya Nadella). Amazon Polly is a text-to-speech engine, and based on this week’s demos, you can expect pretty fluid and natural-sounding utterances. Lex, which is based on Alexa, is a speech-recognition and natural language understanding service. Amazon has had nearly two years and billions of conversations with millions of Amazon Echo customers to train this service.
MyPOV on Rekognition, Polly and Lex: Google and Microsoft introduced similar services earlier this year, but Amazon is right to point out that it has been working on these capabilities for a long time, even if they weren’t available as cloud services. I’ll be really interested to see the degree of training required, particularly for understanding the context of a specific application. AWS demoed an airline flight-booking app with a natural language UI using Lex and Polly. Interpretation was said to based on a built-in knowledge graph that lets you add your own metadata. No doubt developers will have to do more than build an app and plug in the APIs.
Soon enough we will see how numerous and innovative the AI-powered apps are emerging from each of the big public clouds. Google may have an edge on data, between its search engine and the Android operating system. But I’m counting on Amazon’s big edge in cloud developer ranks (not to mention the breadth of its services portfolio) to give it a running start.
Salesforce Einstein: Dream Versus Reality
Oracle Vs. Salesforce on AI: What to Expect When
Tableau Sets Stage For Bigger Analytics Deployments
AWS re:Invent: Five Takeaways On New Services