Devs: with only 64 validator nodes, would an attack by first spamming the network with tx's and then dDOSing a handful of validators not lead to the remaining validators getting overwhelmed and crash? It seems a rather vulnerable setup, esp considering a validator node needs 8GB RAM to handle 5K tx. Even on an expensive VPS, there would not be a lot of room for additional load.
Or am I wrong?
This is definitely a possibility regardless if we are using 64 or 100 validators. Validators need to have sufficient bandwidth, RAM and processing power to handle large amount of transactions and the recent and ongoing RC tests are going to expose these vulnerabilities. With these vulnerabilities exposed our Devs can work out a solution to prevent these problems from happening before we go on mainnet.
I do software, so I know how usually strangers coming in and proposing "simple solutions" is a reason to face-palm, so forgive me if I am talking nonsense...
In such a situation, I would either have a couple of "dark validator nodes" that light up once others go down (latencies and throughput should be measurable) or have a large pool of nodes and have them switch on and off in a random fashion so the mean is 64 active validators.
Both make it harder keeping a sustained dDOS going. In your current setup, I'd be afraid of knock-on effects of a percentage of the validators getting taken out.
