For data forwarded by a Value Stream and stored in an SQL-based persistent database, consider adding a combination of clustered and non-clustered indexing to the database table to enhance read performance, which becomes more important as analytics and visualization scale up, such as with the introduction of ThingWorx Mashups. The exact process and configuration for an ideal table index will vary based on your setup.
A clustered index determines how data in a table is ordered for storage. It is helpful to cluster data in the order that you expect to query it, such as in reverse chronological order.
A non-clustered index creates a mapping to more quickly retrieve data based on a certain property, much like the index in the back of a reference book. However, a non-clustered index does not affect the order in which the data is stored.
Because of the additional processing required for generating and writing the index information, adding indexes to a table can have an effect on the write latency and write data amounts. In our observations, these effects were negligible.