Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Dell ECS: Data Lake with Apache Iceberg
Executive summary Introduction Apache Iceberg introduction Using object storage support Apache Iceberg Case reference Performance report
Overview
Summary References

Overview

Home > Storage > ObjectScale and ECS > Industry Solutions and Verticals > Dell ECS: Data Lake with Apache Iceberg > Overview

Overview

  • People ask the input/output per second (IOPS) size of the Iceberg on the ECS system. Compared to traditional storage systems, object storage systems are built on top of the REST API, and these APIs define different types of discrete I/O operations than other file systems. For example, a storage system that implements a file system in POSIX API semantics has a retention state between I/O operations on the same file. This state keeps a series of logical I/O operations that can be counted discrete. Instead, when using the REST API for object storage, a single API call is used to read the file. The difference between the operation and the calculation makes it impossible to use traditional metrics.

    You can use transactional per second (TPS) and megabyte per second (MB/s) bandwidth to represent ECS performance data, two metrics that describe better the performance parameters of the object storage system.

    EXF900 is an all-flash object storage solution for hyper-converged nodes in the ECS product line for ECS deployments with low latency and high TPSs. According to test results, a single-node S3 10 KB size data write performance can reach more than 10,000 TPSs, and the 100 MB size data can reach more than 1 GB bandwidth. A single-node S3 10 KB size data read performance can reach more than 30,000 TPSs, and the 100 MB size data can reach more than 5 GB bandwidth.