Home > Storage > PowerFlex > White Papers > Dell APEX Block Storage for AWS: MongoDB using Kubernetes and PowerFlex CSI > Test methodology
All the test workloads are carried out using the YCSB standard benchmarking tool. YCSB is popular Java open-source specification and program suite that is developed by Yahoo. This deployment uses YCSB for MongoDB sharded cluster as a benchmarking and performance tool. For more information about YCSB best practices, see Yahoo! Cloud Serving Benchmark (YCSB).
For this white paper, the MongoDB use cases are run on instances with the EBS storage and Instance store as storage types. The performance varies based on the underlying resources, storage volume types, and latency depending on the geographical locations of the instances. To avoid network latency, place the YCSB client pods in the same AZ as the MongoDB data shards. MongoDB is efficient in using CPU and memory. As a result, less stress was placed on the storage subsystem. The following tests show only a small difference between EBS and Instance store storage types. In other deployments, where the datasets are larger (datasets do not fit well in memory), more noticeable performance differences appear.
In each MongoDB data pod, a YCSB database name is created. For varied use cases, different workload types with various parameters such as the number of threads, operation count, and record count were simulated to quantify the cluster performance by observing the total operations per second and the derived latency for that test configuration.
Details about the test configurations include: