Alternate data store for buckets

[eddieantonio]

PartyCrasher has the following main entities:

• Crash reports
• Buckets
• Projects

However, our main datastore, ElasticSearch only directly stores one entity: crash reports; the other entities exist as attributes of crash reports.

This presents a few problems including:

• Not easily being able to do trivial queries (e.g., cardinality) on buckets or projects
• Having to compensate for ElasticSearch's eventually consistent nature in the bucketer. This may lead to bucket splitting (although, eventually consistent insert and more-like-this query would also cause bucketing splitting).
• Having to add each and every threshold as an unanalyzed field in ElasticSearch

Proposal: would an alternate data store (such as an SQL database) be appropriate, or should we continue storing these entities as attributes of crash reports?

[hazelybell]

I think we should continue storing these entities as crash reports, writing the cardinality query for buckets/projects shouldn't be a big deal.

[hazelybell]

So if we need this eventually then we can use a seperate index