5 Questions to Ask Any Streaming Big Data Analytics Vendor
The big data analytics market is rapidly embracing streaming data as the newest source of tremendously valuable time-sensitive insights. Many enterprises are beginning to demand solutions to easily acquire and expose their big data streams for analysis. Last week Forrester released “The Forrester Wave™: Big Data Streaming Analytics Platforms,Q3 2014” (by Mike Gualtieri and Rowan Curran) reviewing a sampling of streaming big data analytics platform options on the market. The report, which doesn’t claim to be an exhaustive search of all vendors, covers the large usual suspects as well as a few specialist vendors. Unfortunately WebAction recently came out of stealth and was not contacted during the research for the report, which is a shame because the WebAction Real-time App Platform is a clear contender to lead the pack (in my humble opinion). Regardless, it is inspiring to see that this huge market is taking shape. Even better that WebAction is philosophically aligned with how the report paints the landscape for streaming big data analytics platforms. (You can purchase the Forrester research report here.)
5 questions to ask your streaming big data analytics vendors in a bake-off
#1 Is the Platform Intuitive To Use and Easy to Extend?
To get maximum value from any technology, it better be easy to use. Your ability to gain timely value from streaming data is limited by how easy it is to access and manipulate your streams. Any great platform should hide the complexity of dealing with streaming big data and let you focus on the business logic. Think about how the data moves through the platform: how it acquired, processed, and delivered. Can you easily add new data sources to your streaming platform, does that include static history and context mined from your data lakes? For processing, is your business logic easy to understand and review? Does it all live in one place, or is the logic and config spread all over the file system? Can you extend the business logic across your streams without having a PhD in Computer Science? When delivering the big data records created from the streams, can you send immediate alerts? Power visualizations? Can you persist actionable big data records to any data store? Will your streaming platform integrate and live in harmony with your existing infrastructure?
#2 How Does the Platform Perform and Scale?
Your platform needs to scale at the pace of your big data streams. We are at the tip of the iceberg with the Internet of Things, as this vision materializes, we will see data velocity, volume, and variety explode. To handle massive uptick in big data streams you need a distributed architecture where streams are partitioned in real-time to flow across nodes. Will your partitioning scheme keep up and keep node cross-talk to a minimum? Scaling out lets you use commodity hardware rather than rely on specialized and expensive big iron.
#3 Can You Correlate Across Streams and Enrich Streams with History and Context?
You probably already have too much of your data kept in silos, you don’t need another ivory tower, you need a way to bring your streaming big data together so that you can watch events unfold across your infrastructure to create immediately actionable insights. Actionability also requires bringing in rich history and context, which is relatively static. This is a key point that seems to be glossed over too many times. If you are going to invest the time and money in a big data streaming analytics platform, you should have one that is flexible enough to tie together events from across your enterprise, including your reference data.
#4 What Time Investment Is Required to Build, Modify, and Deploy Business Logic?
The easier it is to work with the business logic in your streaming platform, the more value you will get from the platform. Many streaming big data analytic platforms on the market take months to install and configure to accomplish a seemingly simple task. Then if you want to modify the business logic, it requires bringing the professional services group back to make the modifications. This rigid approach almost certainly dooms the platform to a single-purpose or academic pursuits rather than keeping up with the ebbs and flows of the business to discover new insights. Business logic for streaming big data should be abstracted back to something that your average business analyst can understand (SQL for example), and should be easy to access and modify through both a command line interface and a modern GUI.
#5 How Does the Streaming Platform Integrate with Existing Systems?
You noticed a correlated event across your big data streams and associated that back to one of your most prized customers, now what do you do? Your platform should have more than one way to integrate across your enterprise. Can you send alerts and trigger workflows from the platform? Can your in-memory big data records be written to existing data stores for other systems to easily consume?
These are just a few things that are top of mind when considering streaming big data analytics. Some other concepts to explore as you continue your discovery about streaming analytics: is the platform fault-tolerant? does the platform offer health monitoring and dashboards? and how is error traceability managed?
One final thought, while your real-time streaming analytics platform should live in harmony with your existing batch architecture, having a single end-to-end solution makes the whole process easier to manage. As you probably have guessed, the WebAction Real-time App Platform is all of this and more. Schedule a demo to see it in action.