Full Archival node with indexer for extensive queries optimum configuration

We need to create an full archival node with indexer fir extensive query.
What is the optimum configurations. Can this node be a consensus participating node?

Welcome to Algorand!

I would start from the following configuration and upgrade if needed:

  • 4 vCPU (you may start with 2 vCPU if you really want to cut costs)
  • 16GB RAM (don’t compromise on the RAM)
  • 1TB SSD (no need for local SSD, but cloud persistent HDD would not be sufficient; for example, on GCP, “balanced persistent disks” are a minimum requirement)
  • 100 Mbps connection

The above should allow you to run algod in archival mode + PostGres + indexer without any issue today.
But I do not know what load it would support.
You would need to do specific tests for your specific workload.
Furthermore as Algorand gets more and more adopted, the specs may need to increase.
This answer was written mid-April 2021.

You can definitely run a participation node on an archival node.
If you don’t have a huge amount of stake (say less than 1M Algos), I do not see any issue.
For a huge amount of stake, running a separate participation node would allow you to achieve a slightly higher level of security, by not opening any port on the server (and hence reducing the attack surface).
Note that you should never store your spending key on the participation node, so the worst that can happen is losing your participation key, but this can have a negative impact on the Algorand blockchain if you have a huge stake.

Hi Fabrice,

Thanks for the prompt reply.

Lim Tai Wah
Eastern Blu Services PTE Ltd
Singapore 218418