This document outlines the concept of a KORD indexer, a KORD service which provides efficient access to data stored within the KORD network.
The KORD protocol is at an early stage of development, expect this document to change frequently during this early phase.
As outlined in the KORD Stream concept document, the presence of objects within the KORD network are communicated via persistent, append-only streams. If you want to find a particular object in the network which you know has been communicated in a particular stream, then you have the option of incrementally reading through that stream until you find that object.
Due to the append-only nature of streams however, this task will get ever less efficient over time as the stream grows in size. What you need is an index which acts as a lookup table into the stream, so when looking for an object with a particular property value, you instead retrieve a link to the object directly from the the index object.
A KORD indexer provides such an index object by reading objects from a stream and assigning particular property values of those objects as properties of the index object which link to those objects.
As a concrete example, imagine a stream of objects representing 1,000 people,
and you want to query them based on gender. The indexer would construct a
gender index object with two properties male and female, the values of
which would point at two KORD streams, one containing just males and the other
just females.
TODO: outline incentives (similar to aggregator)
TODO: outline accountability (similar to aggregator)