APPA

In traditional data fabrics, messaging services such as Apache Kafka are used to pass data between producers and consumers. Queues and topics both enable pub/sub communications, but both processes are bandwidth-inefficient, as data targeted for multiple consumers is duplicated and transmitted repeatedly, and some transmitted data may be immediately dropped due to a lack of need by the consumer, further wasting bandwidth.  

APPA brings intelligence into the data distribution layer, offering familiar pub/sub and REST models while adding additional hooks to further optimize transactions. This comes in the form of consumer-specified data request definitions, where systems define what data they want to see from their data source(s), such as specific metadata values or compound triggers across different elements. This enables APPA to identify on receipt from the consumer if a given producer is interested in the data, saving both transmission time and resource requirements otherwise spent discarding the data at its destination. This is particularly beneficial in edge processing, where APPA at the edge is made proactively aware of potential consumers and can determine not to transmit data with no destination, optimizing what may otherwise be a soda-straw uplink.  

As part of requesting data, consumers can also request a desired format as part of their subscription, abstracting the producer’s output with their own system’s needs. If the provided data does not match this format, APPA automatically inserts a translator to map data transparently and within the distribution stream, saving time if multiple systems require the same format and decoupling individual components from integrating directly. These translators can be built by Soresu or by individual vendors, with the latter added to the APPA translator library to share across systems after being tested and validated for compliance with data standards.