- StreamBase LiveView
- StreamBase CEP
- StreamBase Adapters
- Use Cases
- News & Events
Frequently Asked Questions
Why did we build StreamBase LiveView?
We built StreamBase LiveView because firms with event-oriented infrastructure (message buses, FIX messages, social media) came to us because StreamBase is all about processing information in motion, rather than information at rest. Traditional disk-based BI simply was a non-starter for fast-moving environments that require automation, like trading or e-commerce. One client in particular, a major international bank, partnered with us to build and deploy the first version of StreamBase LiveView, and we subsequently commercialized that product and released it as StreamBase LiveView.
How is StreamBase LiveView different than traditional analytics?
First of all, StreamBase LiveView is really comparable to two main architectural elements that you might be familiar with: it’s part data warehouse, and part real-time analytics. The StreamBase LiveView server manages client connections and queries, continuously re-evaluates queries in real-time, and manages input data streams from any of the over 150 connectivity options in the StreamBase catalog. The StreamBase LiveView Desktop provides an end-user query mechanism, notifications, graphing, charting, heat maps and other presentation interfaces so that end users can slice and dice streaming information however they choose.
So StreamBase LiveView is faster. So what?
Real-time analytics isn’t about speed; it’s about a new interaction model. Mobile technology has taught us all about a new style of managing our information: push. Alerts from our calendar, Twitter feeds, and Facebook friends are now natively pushed in modern computing technology like iOS clients, whose iOS5 interface has revolutionized the way we all interact with our data – we expect our devices to tell us when something has changed, rather than having to ask a computer over and over and over again.
So real-time analytics is as much about the interaction style with users – push – rather than raw speed (which is also a differentiation).
In other words, we now have a live view of our social networks, why shouldn’t we have a live view of enterprise data as well?
Updates to the client can’t really be “live,” can they?
Clearly, GUI technology and messaging needs to be carefully configured, and the definition of “real-time” is relative in relation to how quickly screens can be refreshed. But, in theory, yes, there can be merely milliseconds (some would call this “near real-time”) between the time an event occurs, the time the StreamBase LiveView server processes the query, and the time the results are pushed to end users.
Aren’t high-speed data warehouses as close to real-time as most applications require?
No, and it’s more than just about speed, it’s also about the fidelity or detail of data.
First, let’s talk about pure speed. Data warehouses are designed for historical data, and even though many modern warehouses can load data quickly, they load digested data, often from a complex labyrinth of batch processes, ETL, or Map/Reduce jobs in Hadoop to load data. Once loaded, sure, data warehouses can have acceptable response times, often on an order of magnitude of about between 15 minutes and an hour. For many, that’s fast enough. But the world of data is changing, and thanks to the ability to monitor social data, electronic trading data via FIX, or sensor-driven information – the near real-time decisions that need to be made on this type of data can’t wait an hour.
Second, there’s fidelity. In order to make up for the computing and storage overhead associated with file-oriented data warehousing, traditional tools simply drop out lots of data. For example, the purpose of Map/Reduce is to reduce the information digested by a warehouse. Because StreamBase LiveView operates on data before it’s digested by the warehouse, it operates on all streaming data, and can spot patterns in data streams that normally would simply go undetected, because that data would have been aggregated and lost by the time it reaches the warehouse.