Most Ethereum nodes run on hosting and AWS, and that’s not okay.
More than six thousand (or 56%) of all Ethereum nodes are running on hosting providers, 2,500 (or 36.%) of those being on Amazon Web Services (AWS). The runner up for the most Ethereum nodes is residential, clocking in at a calm 4,500 (or 38.91%).
This is great! People are running a Ethereum nodes in their home, without worrying about AWS (although they are good at it) going down or some other hosting provider while being used for a dapp and may impact user performance (for example, two of the most popular hosting services, alchemy, just suffered a down yesterday, impacting Zerion for about one day (and others), Infura, the #1 node-as-a-service provider, has had many outages. Just search up “Infura down”.) This poses a reliance on “hosting” nodes (including alchemy, quicknode, and infura), as if just AWS goes down (it’s very unlikely) the network would be in a interesting state. Most mining pools/nodes are connected together, so blocks would still go through, although not many transactions would pass. To someone in the “Residential” side, they would not see any difference. (Except for some lower gas prices :) )
So are there any solutions? In theory, everyone could just dump AWS and other hosting providers such as Infura and the blockchain would be much much harder to attack. The reality? We NEED hosting providers like Infura and AWS. Having 647gb of fast storage (at least on time of writing) just to sync the chain (fast) and then adding 1gb every day just isn’t practical for the average person. There is also almost no economic incentive to running a node, although projects like vipnode are working on this problem. You could run a node during ICO times (gone now though) when infura would get clogged to get a tx out. Running a node like this on AWS should cost you ~60 USD per month. That’s no small amount for a node that has no incentive to it, bringing people to Infura: cheap. works.
But Tennis, what will happen with Eth 2.0? Well for the time being to be a validator you need to have a eth 1 node (and the struggles of syncing that fatty chain) as well as a eth 2 node (quite small right now at 83gb for a archive node), but that will get longer as the chain runs for longer and longer. Eth 2 is meant to be the permanent solution to eth and syncing it will become more and more of a struggle as we go on. Some people run their eth2 validators with hosting providers such as AllNodes, same problem as Infura. Running a validator with Infura as your eth1 node and you running your own eth2 node is diminishing the risk, yes, but isn’t making you trustless, and remember: you have your eth at stake. So take action. Stop using hosting or nodes-as-a-service providers such as Infura, and run your own. You are helping the chain as well as yourself.
All nodes data including size of chain and amount of nodes on AWS is provided by Etherscan (for sizes) and EtherNodes (everything else). Both are great tools and I recommend them both.
