Amazon Web Services (AWS) is one of the most popular providers of public cloud services. Running databases on AWS is an essential part of using its cloud services. But how can we best leverage the cloud resources and have the database run at its best performance?
This post demonstrates how we at PingCAP deploy and optimize TiDB for production on AWS Cloud. It’s our hope that the recommendations provided here will also help you configure the TiDB service most suitable for your own workload.
To evaluate the performance of different configurations, we ran TPC-C and Yahoo! Cloud Serving Benchmark (YCSB) tests on a typical three-node cluster. TPC-C is a common benchmark standard that simulates an intensive online transaction workload, while YCSB covers the most common cloud application scenarios. This article discusses the results in detail.
When an application is built on top of a persistent database like TiDB, the database eventually translates user interactions into reads and writes on disk storage. This is why this article focuses on how to better utilize AWS cloud storage as a new storage medium for TiDB.
In this post, we consider using AWS Elastic Block Storage (EBS) as the primary storage for the TiDB service. Unlike traditional local disks, EBS is:
Even though EBS has a smaller annual failure rate (AFR) than commodity hardware, it’s still not reliable enough for business-critical applications. They require cross-region availability, which is only possible with distributed databases replicated in multiple availability zones (AZ). In this regard, TiDB is a good match for EBS.
#tidb #aws #lower cost