Skip to main content

Creating a cluster

To create your first ClickHouse cluster, click on ‘create cluster’ and you will be presented with all the customization options for deploying your cluster:

Cluster name

You may name your cluster any name (including spaces, numbers and special characters). With a maximum of 50 characters.

Cloud provider

You may deploy your cluster on either AWS or GCP.

This can be directly through Gigapipe (advised) or 'self-hosted' on your own AWS/GCP account.

NOTE: Self-hosted option will require you to deploy Launchpad yourselves. Please contact support@gigapipe.com for more info.

For more information about AWS please click here: https://docs.aws.amazon.com/AmazonECS/latest/userguide/clusters.html

For more information about GCP please click here: https://cloud.google.com/compute

Region

To obtain the lowest latency working with your data, it is recommended that you choose the region closest to where you work. Unless you have specific requirements to have your data in a specific region.

All geographical regions supported by AWS and GCP are available in Gigapipe. If your preferred region is not in the existing list, please email support@gigapipe.com and request the new region to be added to the system. It will usually be available within 24 hours.

List of AWS regions: https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Concepts.RegionsAndAvailabilityZones.html#Concepts.RegionsAndAvailabilityZones.Regions

List of GCP regions: https://cloud.google.com/compute/docs/regions-zones

Regions are grouped by continental area. You will only be able to see regions available for the specific cloud provider selected (AWS or GCP).

Gigapipe will NEVER move your data from your specified region. All local data laws and compliance apply to your specified region.

Machine

Gigapipe recommends and supports machine types best optimized for deploying your ClickHouse clusters. These are:

Shards and Replicas

You must decide how many Shards (principal nodes), and Replicas (mirrors), to provide the desired performance and contingency infrastructure for your installation.

You can deploy on as many or as few shards and replicas as needed. We would recommend single shard single replica (1 node) for testing, and 3 shards one replica (3 nodes) for any production use to ensure high availability.

Number of nodes = Shards * Replicas

For more information on nodes, shards and replicas please click here: https://en.wikipedia.org/wiki/Shard_(database_architecture).

Disks

When deploying ClickHouse through Gigapipe you can have multiple disks of multiple available disk types.

Disk types available:

  • AWS
    • gp2
      • General purpose SSD
      • Low-latency interactive apps
      • Development and test environments
    • io2
      • Provisioned IOPS SSD
      • Sub-millisecond latency
      • I/O-intensive database workloads
    • st1
      • Throughput Optimized HDD
      • Data warehousing (for less frequently queried data)
      • Minimum 150 GiB per node

For more information on AWS available disk types please click here: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-volume-types.html

  • GCP
    • (SSD) - Regional SSD PD
      • Fast and reliable block storage with synchronous replication across two zones in a region
    • (HDD) - Regional standard PD
      • Efficient and reliable block storage with synchronous replication across two zones in a region

For more information on GCP disk types please click here: https://cloud.google.com/compute/docs/disks

You can deploy using a single disk however if you intend to use storage policies or TTLs to offload less frequently queried or older data, then we advise deploying larger and slower less expensive disks alongside the primary disk.

Admin username/password

Name your ClickHouse cluster ‘Admin username’ and ‘Admin password’ (Confirm password). You’ll need these credentials if you intend to directly interact with your ClickHouse cluster without using the Gigapipe UI.

Pricing calculator

All disk costs are passed along at cost (no mark-up) so you are never punished for storing more data!

Clusters are billed on an minute by minute basis; the monthly estimate is the expected cost if the cluster were to be running from an entire month (a 730 hour month).

Create Cluster

After reviewing all the details and the estimated price for the monthly run of your new cluster, click on 'Create Cluster' button to deploy your cluster in your specified region/provider.