Sizing – choosing the right machine to do the job

Sizing is based on the number of concurrent users and the number of concurrent queries that you have. For <10 concurrent users and <100 active queries, a tiny machine that has a dual core, 2 GB memory should be enough.

On the other hand, if you have 50-100 concurrent users and 1,000 active queries, you will need at least 16-32 GB RAM and a quad-core CPU.

• Redis - Memory intensive 
• Postgres - CPU and memory
• Celery - CPU and memory (the more workers you have, the more CPU you need. If workers won't keep up and build up a queue, it starts to consume memory)

If you're hosted in the cloud, you can always start with a smaller machine and expand if needed. The same goes for deploying Redash inside a VM in your own data center.

As a good starting point, you can use the following instance types on GCE/AWS:

• On AWS: t2.medium (2 vCPU, 4 GB RAM)
• On GCE: n1-standard-1/ n1-standard-2 (1/2 vCPU and 4/8 GB RAM accordingly) 

For more intensive usage:

• On AWS: m4.2xlarge/m5.2xlarge (both with 8vCPU, 32 GB RAM)
• On GCE: n1-standard-8 (8vCPU, 32 RAM)