Good to know about the partial indexes on Postgres, but I am little queries to ask you as an analytics company your system is write-heavy system, how did you justify increasing latency of writes, I am pretty sure that you have some kind of queue in front of Postgres to overcome this problem, secondly what I understand from the blog and nature of event-based analytics it's generally time series data, What made you stick to Postgres where I can easily see lot of different datastores build specifically to deal with time series data like Elasticsearch and influxDB ? can you give me some insights on why Postgres.