Inbound Processing

StreamBase processes data on the fly to deliver results in milliseconds. It offers unparalleled performance for time-sensitive organizations where "right now" has to be, literally, RIGHT NOW.

When solving a real-time processing or analytic problem, many organizations familiar with relational databases (RDBMS) attempt to "shoehorn" a continuous query into the prevailing RDBMS model in which incoming data is first stored and then indexed in a disk-based RDBMS. Once the data is stored and indexed, the querying and processing tasks may be implemented in Structured Query Language (SQL). This approach is called outbound query processing, where queries are run after the streaming data is indexed and stored.

In streaming data applications, by contrast, data records flow through the query and application processing steps, which may in turn transform this data while it is moving. This highly-efficient "process before (or instead of) store" model is called inbound query processing. Running such "continuous" SQL-like queries is well suited to many applications in the financial services world where operating on time-series data is central to trading, pricing, and compliance activities.

First let's illustrate the older outbound processing model, where data is indexed and stored before queries are executed and results delivered:

Now let's look at the inbound processing model, which provides results delivered in-flight, as produced, with low latency:

With inbound processing, queries are executed on streaming data over specified windows of time before it is placed in storage. Indeed, because not all data need be stored, inbound processing does not incur the significant overheads of indexing, storage, and retrieval. Of course, many applications require that some or all of the streaming data be stored for subsequent retrieval (for example, to meet archival or audit requirements). Even here, doing inbound processing has significant advantages, because it overlaps the processes of stream processing and historical storage in time.