Sharedstake follows a LEAN startup buy over build methodology and leverages cloud services with high performance and uptime for our ETH2 nodes.
To hedge against the risk of synchronized downtime with the majority of cloud hosted ETH2 nodes being on AWS or GCP, we use a German Cloud provider.
We currently run a cloud server with 16GB RAM, 8 cores, 150GB Disk space and take monthly backups, and plan to do quarterly resets wherein we sync the beaconDB from scratch to combat bloat we have observed in long running Prysm instances.
For ETH1 data we use Alchemy APIs with a fallback on Infura.
We use Prysm for our nodes due to:
Several benchmarks demonstrating superior performance
Benchmarks demonstrating less slashing risk and historic slashing incidents than other Validator software
Use of Golang which our engineers are familiar with
Web UI that we can screenshot for our users for transparency
An active and helpful community and discord
We've additionally ran our own benchmarks and have tips for others who would choose to run prysm and characteristics of long running Prysm nodes included in the post mortem from an incident where our nodes ran out of disk space. You can read it here:
GitHub - chimera-defi/eth2-quickstart: Scripts to get a eth2 merge ready node setup in secondsGitHub