GCP configuration options

ThoughtSpot can be deployed using several different GCP instance types. ThoughtSpot can be deployed in your GCP environment by deploying compute (VM) instances in your VPC, as well as an underlying persistent storage infrastructure. Our recommendations for instance types differ, depending on whether you are accessing your data by loading it into ThoughtSpot’s in-memory database, or if you are connecting to your data in a cloud data warehouse. Currently two configuration modes are supported by ThoughtSpot:

Mode 1

Compute VMs

SSD Persistent Disk storage-only

Mode 2

Compute VMs

SSD Persistent Disk and Google Cloud Storage (GCS).

For more information about Persistent Storage, see Zonal Persistent SSD disks in Google’s Cloud documentation.

For more information about Google Cloud Storage, see Cloud Storage Buckets in Google’s Cloud documentation.

All GCP VMs (nodes) in a ThoughtSpot cluster must be in the same zone (and, therefore, also in the same region). ThoughtSpot does not support deploying VMs (nodes) of the same cluster across different zones. For more information, see Regions and Zones in Google’s Cloud documentation.

ThoughtSpot GCP instance types

When choosing an instance type, ensure that it uses Intel CPUs. Our recommendations for instance types differ, depending on whether you are accessing your data by loading it into ThoughtSpot’s in-memory database, or if you are connecting to your data in a cloud data warehouse.

Instance types for in-memory data

VMs with Persistent Disk-only storage

persistent storage ssd
Persistent Disk storage: recommended capacity and volume, based on user data capacity, on a "per VM" basis
User data capacity[1] Instance type CPU/RAM Zonal Persistent SSD Disk volume[1] #Rows[2] Required boot volume

208 GB

n1-highmem-64

64/416

2x 1 TB

> 4 billion

200 GB for each node

312 GB

n1-highmem-96

96/624

2x 1.5 TB

 — 

200 GB for each node

100 GB

n1-highmem-32

32/208

2X 400 GB

1 - 4 billion

200 GB for each node

20 GB

n1-highmem-16

16/122

2X 400 GB

< 1 billion

200 GB for each node

180 GB

n1-standard-96

96/360

2X 1 TB

 — 

200 GB for each node

VMs with Persistent Disk and Google Cloud storage

persistent storage ssd gcs
Persistent Disk and Google Cloud storage: recommended capacity and volume, based on user data capacity, on a "per VM" basis
User data capacity[1] Instance type CPU/RAM Zonal Persistent SSD Disk volume:pvm[] #Rows[2] Required boot volume

208 GB

n1-highmem-64

64/416

1X 500 GB

> 4 billion

200 GB for each node

312 GB

n1-highmem-96

96/624

1X 500 GB

 — 

200 GB for each node

100 GB

n1-highmem-32

32/208

1X 500 GB

1 - 4 billion

200 GB for each node

20 GB

n1-highmem-16

16/122

1X 500 GB

< 1 billion

200 GB for each node

180 GB

n1-standard-96

96/360

1X 500 GB

 — 

200 GB for each node

For most instances, the per VM recommended user data capacity is set at 50% of the available RAM on the instance. However, in the case of our 16CPU/122GB RAM and 32CPU/208GB RAM instances, we support user data sizes below those numbers to budget for application overhead.

Instance types for cloud data warehouse connections

VMs with Persistent Disk-only storage

Persistent Disk storage: recommended capacity and volume, based on user data capacity, on a "per VM" basis
User data capacity[1] Instance type CPU/RAM Zonal Persistent SSD Disk volume:pvm[] Required boot volume

Up to 1B rows

n1-highmem-16

16/122

2X 400 GB

200 GB for each node

Up to 4B rows

n1-highmem-32

32/208

2X 400 GB

200 GB for each node

4B+ rows

n1-highmem-64

64/416

2x 1 TB

200 GB for each node

VMs with Persistent Disk and Google Cloud storage

Persistent Disk and Google Cloud storage: recommended capacity and volume, based on user data capacity, on a "per VM" basis
User data capacity[1] Instance type CPU/RAM Zonal Persistent SSD Disk volume[1] Required boot volume

Up to 1B rows

n1-highmem-16

16/122

1X 500 GB

200 GB for each node

Up to 4B rows

n1-highmem-32

32/208

1X 500 GB

200 GB for each node

4B+ rows

n1-highmem-64

64/416

1X 500 GB

200 GB for each node


1. per VM
2. Recommended for connected tables