I'm considering replacing or supplementing AWS NAT Gateways with FCK-NAT to reduce costs, especially since our workloads move several terabytes of traffic each month. Has anyone operated it in production, and what reliability, maintenance, or scaling issues should I plan for?
3 Answers
I run one FCK-NAT instance in each of my three Availability Zones, so every zone has an independent path for outbound traffic. It has worked well enough that I rarely need to think about it. For a smaller company, this setup can be a good cost-saving option as long as you are comfortable owning the operational maintenance.
I've been running FCK-NAT for about six months and haven't encountered any problems. So far it has been stable and has performed as expected in production.
Good to hear. Did you have to handle any unexpected maintenance or recovery issues during those six months?
I've used FCK-NAT in several production clusters for around 18 months, and it has been remarkably trouble-free. The savings become significant when you're pushing multiple terabytes of traffic each month. The main tradeoff is that you have to maintain the instance yourself, and an outage can interrupt outbound connectivity until it recovers. Make sure automated recovery and startup configuration are tested.
Would running two instances across separate Availability Zones with an Auto Scaling Group cover the availability concern? We currently handle more than 5 TB of traffic.
That sounds like a solid approach. Two instances in separate Availability Zones with an Auto Scaling Group may also work for us, though we would need to verify throughput and failover behavior at our current traffic volume.