[FEATURE] Allow variations of the workload(s) with compatible data types

[CaptainDredge]

Is your feature request related to a problem?

OpenSearch supports several data types such as byte, short, float, double, geopoint and so on. Backend performance of operations on these data types can change, for instance as a consequence of sort optimizations. However, the existing workloads don’t exercise all these data types in the index mapping even though some of them are compatible with one another for eg. long and unsigned_long, strict_date_optional_time and YYYY-MM-DDTHH:mm:ssZ etc. As an OpenSearch developer, I want to be able to track indexing performance for operations on compatible data types. Right now one has to manually change index.json for changing data types in the mapping.

What solution would you like?

One should also have the option to provide type of any field defined in workload mapping as workload-params as being done for a specific type in this PR

What alternatives have you considered?

Continue with ad-hoc index mapping modifications

[CaptainDredge]

@reta any thoughts on this one? I don't see a clean way to implement it just by jinja template magic

[reta]

@CaptainDredge thanks for bringing this one up, strictly speaking I don't like this solution (#86) but as of today, this is the most straightforward way to alter the workload without copying tons of JSON.

@reta any thoughts on this one? I don't see a clean way to implement it just by jinja template magic

Thinking about that, I actually came to the same kind of conclusion: would be great to have a "flavor" of the index (template) geared toward different data types. It will surely work only for compatible data types but I think that:

• templating the index
• introducing the variations ("flavor" like unsigned_long, scaled_float, ...)

would be a much cleaner approach.

@IanHoang @gkamat do you have an option folks?