We use azure SQL (basically mssql) to perform something similar -- this might be a more apt comparison. We deal with largish amounts of JSON event data, and un

We use azure SQL (basically mssql) to perform something similar -- this might be a more apt comparison. We deal with largish amounts of JSON event data, and unpack specific fields to computed columns. This allows to use clustered indices, standard analytical tools, etc., while leaving the raw data as it is. Performance is mostly fine for analytics (time series aggregation and such). It is also a very flexible approach.