Towards Performance Evaluation of Cloudant for Customer Representative Workloads

摘要

NoSQL Database Management Systems (DBMS) have increased in prominence and market share in the last few years. IBM Cloudant is one of the well-known enterprise NoSQL DBMS which has been offered in both cloud and also as an onpremise standalone product (i.e. Cloudant Local). Cloudant has a large number of customers such as Samsung, Adobe, DHL, etc. In this paper, we will demonstrate how Cloudant can maintain its high throughput for customer-representative workloads which have a mix of create/read/update/delete operations and also queries that contain views. Moreover, we also show how we can dynamically add/remove database nodes to/from the cluster without bringing down the working nodes. We also demonstrate how we have improved the performance for queries with views by changing a native library.