This section broadly describes how we installed, configured, and tested SQL Server 2019 Big Data Cluster. Rather than including step-by-step instructions, we focus on the key elements of our successful Big Data Cluster implementation and provide links to documents that detail the installation requirements.
Based on the working assumption that virtualization is an existing part of the data center, we do not address VMware vSphere installation in this section. Many of the steps apply to bare metal and virtualization; thus, if the organization is considering bare metal, the steps still apply.
The high-level steps for installing, configuring, and testing this solution are as follows. The subsequent sections provide additional details.
- Install Docker, Kubernetes, and the VMWare vSphere CSI plug-in:
- Install the Docker Enterprise Edition in a VM on the VxRail system. The operating system within the VM is Red Hat Enterprise Linux.
- Install and configure a local private Docker registry.
- Install Kubernetes.
- Install and configure the vSphere CSI plug-in.
Note: Container services exist as instance images that reside in registries. Docker supports its own public registry service called the Docker Hub, which holds more than 2.7 million container applications, including Redis, MongoDB, and MySQL instances, which are all freely available. Similarly, Microsoft has its own Docker registry (Microsoft Container Registry), which holds all the images that are required for Big Data Cluster deployment.
- Deploy Big Data Cluster:
- Pull the most recent SQL Server 2019 Big Data Cluster for Linux container image from the Microsoft Container Registry.
- Push the SQL Server 2019 Big Data Cluster to the image in the local private registry.
- Configure vSphere dynamic storage provisioning.
- Install Big Data Cluster.
- Import the TPC-H data into Big Data Cluster.
- Test data virtualization.
- Migrate data to the data pool by ingesting data from the stand-alone SQL Server database.
- Ingest 10 TB of TPC-H data into Big Data Cluster.