Subclusters
Accelerator uses subclusters, which allows you flexibility in separating and managing data workloads. By default, all Accelerator databases contain a single primary subcluster; you will often want to create additional subclusters to separate and manage workloads.
Create and Manage Subclusters
All Accelerator databases contain a single defaut primary subcluster named default_subcluster. This subcluster is created during the creation of a new database and contains all the nodes issued during the creation process. You will often want to create subclusters to separate and manage workloads.
Note: In addition to the one primary subcluster, Vertica Accelerator databases support a maximum of three secondary subclusters.
Every subcluster deployed in a database can be of a different EC2 instance type, providing increased flexibility. Read more about the different EC2 instance types on the Create and Connect page. To avoid latency issues, secondary subclusters are co-located at the same region and availability zone as the primary subcluster of that database.
Create Secondary Subclusters
To create a secondary subcluster, ensure the database is up and then navigate to the Accelerator Home page. In the bottom left corner of your database select Add Subcluster. Choose the name, cluster size, and EC2 instance type for the subcluster. Note that once the subcluster has been created, its name cannot be changed. The size and EC2 instance type can be changed.
Accelerator currently supports the cluster sizes Small, Medium, and Large. The supported EC2 instance types are i3.2xlarge, i3.4xlarge, i3.8xlarge, and i3.16xlarge. This instance family is storage-optimized for low latency and delivers high IOPS at a low cost. To learn more, visit AWS EC2 i3 Instances.
Start and Stop Subcluster
Subclusters can be started and stopped as needed to accommodate changing workloads and minimize costs. You can automate starts and stops using Schedule-based autoscaling or Elastic Autoscaling.
Note: In order to start a secondary subcluster, the primary subcluster (default_subcluster) must be up as well.
Stopping the primary subcluster (default_subcluster) will also stop all the secondary subclusters.
To start or stop a subcluster, locate the subcluter on the Accelerator home page and select either Start or Stop. Note that stopping a subcluster will cancel all scheduled jobs and running sessions.
Once a start or stop action has been initiated, it may take several minutes for AWS to execute the request. Select Refresh at the top-right corner of the database to get the latest subcluster status.
Connect To Subcluster
You can distribute the workload among subclusters as per your preference. To do so, you will need to first connect to that subcluster.
If you have configured your database’s autoscaling policy to Elastic Autoscaling, you will have to connect to the autoscaler DNS.
To connect to a subcluster, navigate to the Accelerator Home page, and next to the subcluster you would like to connect, select Connection Info. Select your preferred connection method. Connecting with Vertica SQL (vSQL) is optimal for managing and administrating the subcluster; however, you can also connect via another programing language, allowing for greater customization. In order to connect another programming language, you must choose your preferred operating system, download and install the client driver as needed and specified on-screen, and after installation, return to the Accelerator screen to complete instructions for connecting. Or, you can use a preferred Partner Tool: see the full list of Vertica’s Technology Partner Integrations and follow their respective connection string on your Command Prompt.
Troubleshooting: Forced Start and Stop
Sometimes AWS runs into unforeseen error conditions after a Start or Stop action has been initiated. Such conditions often result in the action being incomplete and the subcluster status on Accelerator remaining in the ‘Starting’ or ‘Stopping’ state for an extended period. If this happens, you should try selecting the ‘Forced Start’ or ‘Forced Stop’ option. This re-initiates the Start or Stop action with AWS. To execute this option, navigate to the Home page and locate the subcluster you are troubleshooting. From there, select the dropdown button next to either the ‘Starting’ or ‘Stopping’ state. From the dropdown, select either the Forced Start or Forced Stop option.
Resize Subcluster
You can resize your subcluster to accommodate increased or decreased workloads. Navigate to the Home page and find the subcluster to be resized. From there, expand the More menu () on the top right corner of the subcluster and select Resize. Choose a new cluster size and select Save and Deploy.
Edit EC2 Instance Type
You can edit your subcluster’s EC2 instance type to better optimize your workloads. To edit the instance type of any subcluster, the primary subcluster must be up. If you are editing the instance type of a secondary subcluster, that secondary subcluster must be down.
To edit the EC2 instance type, navigate to the Home page and locate the subcluster with the instance type to be edited. From there, expand the More menu () on the top right corner of the subcluster and select Edit Instance Type. Choose a new instance type and select Save and Deploy.
Drop Subcluster
Dropping a subcluster will delete it and free all associated nodes.
Note: The primary subcluster cannot be dropped. Instead, you can terminate or delete the database.
To drop a subcluster, both the primary subcluster in the database and the secondary subcluster to be dropped must be up. Elastic autoscaling, schedule-based autoscaling, and idle shutdown must all be disabled before dropping a secondary subcluster.
To drop a subcluster, navigate to the Home page and locate the subcluster to drop. From there, expand the More menu () on the top right corner of the subcluster and select Drop subcluster.