bg
Stream – BangDB = NoSQL + AI + Stream

Stream in BangDB

Stream is a concept which is available in all flavours of BangDB i.e. embedded, Server and p2p.

However, minor behavioural changes could be there with respect to the flavour of the db, but more than 95% of things related to stream would remain same across all different BangDB flavour.

Stream processing

The meaning of stream processing in BangDB is the handling of set of events or data which is time sensitive or timeseries in nature. The handling of events start from ingesting these data into the system, then processing them for various dimensions, training or predicting on these data and finally taking some action as required. Finally storing the data for further analysis or BI. Visualisation is kind of associated with it, but not required from core processing perspective.

This whole process for beginning to end can be cumbersome and difficult to put together and there are several reasons for the same, starting from ingestion to processing (batch or otherwise) to model training or prediction to finding interesting patterns to taking action to storing data for BI or visualisation. In many cases, developer has to take different components from different vendors/open sources, then stitch them and then write some application on top of it. Even if there are some basic framework available, it's hard to deal with them due to time to get started, to be able to process every single event to avoid batch processing to manage and scale the system well and finally to enable newer advanced use cases in continuous manner.
Moreover, in many cases upfront structuring of data is very difficult and then being flexible to changes in the data structure at run time is even more complex as they tend to break the processing.

BangDB has designed the streaming system keeping all of these in mind. Focus has been on to enable developer to quickly start with the problem at hand without spending time and resources on stitchin a platform, then writing code for parsing data, then write code to process data then to train model somewhere else and deploy them in the pipeline, then store data somewhere else, then plug in visualisation etc. With BangDB developer can get started in very simple yet powerful manner.