Gateway scaling and high availability

[kralicky]

• Ensure gateway can be properly scaled without issues
• Investigate whether autoscaling would be useful
• Perform load testing with large numbers of downstream clusters

Questions to consider:

• How do agents reconnect and upgrade when the gateway gets upgraded
• Distributed locking for plugins?
• Shared cache for plugins?
• Jitter/Backoff for agent upgrades and connections

Need to do load testing of gateway. Look into load testing automation.