The MapReduce workflow starts to process experiment data nightly when data of the previous day is copied over from Kafka. At this time, all the raw log requests are transformed into meaningful experiment results and in-depth analysis. To populate experiment data for the dashboard, we have around 50 jobs running to do all the calculations and transforms of data.

We've always counted on SQL Server as our database backend. It has served us well over the years. It isn't the cheapest part of our stack, but with the plethora of tools provided by 3rd parties, we have found an incredible and scalable method of keeping our data available and easy to maintain.

in 2009 we open sourced mrjob, which allows any engineer to write a MapReduce job without contending for resources. We’re only limited by the amount of machines in an Amazon data center (which is an issue we’ve rarely encountered).

The massive volume of discovery data that powers Pinterest and enables people to save Pins, create boards and follow other users, is generated through daily Hadoop jobs...

Our core systems that we integrate with are using SQL Server 2012 / 2016 database servers. We use database views on core system databases to help build our domain model.

Main transactional database. SQL Server 2012 Enterprise with AlwaysOn Availability Groups for high availability and disaster recovery.

Importing/Exporting data, interpreting results. Possible integration with SAS

TBD. Good to have I think. Analytics on loads of data, recommendations?

Multi-model DB: document (JSON), graph, key-value

TO BE EVALUATED

Managing script output and input, as well as data cleansing.

Persist user actions and information in documents

사용자 팔로잉 관계 데이터 저장