Source: Observatorioblockchain / Jose Maldonado
One of the serious problems of Ethereum is its increasing dependence on centralized cloud services for its operation, such as Amazon or Google. The situation is so serious that Amazon Web Services alone accommodates more than 53% of all Ethereum nodes that are currently active in hosting mode. This means a value greater than 30% of the global nodes of the network.
Ethereum is relying more and more on centralized clouds, but Ethereum Core doesn’t seem to care too much. Even knowing the enormous problem that this means: it is enough for Amazon to want to ban them to shoot down more than half of the Ethereum nodes and make it vulnerable. In fact, the distribution of Ethereum nodes in hosting (represents 67.95% of the entire network), is as follows:
Ethereum, Amazon and decentralization
All of those companies run the majority of the Ethereum network. While the residential distribution (represents 29.71%) is as follows:
This data, available on Ethernodes , underlines that the centralization of Ethereum is a serious problem that continues to grow more and more.
An increasingly serious situation with no close solution
What is happening in Ethereum and its nodes is the sum of several problems that have been accumulating continuously and that until now do not have an accessible solution. First of all, Ethereum nodes are extremely hungry for storage resources. For example, the official GETH node requires a minimum of 2 TB of storage space in order to create an Ethereum full node. This is almost three times what is required for a full Bitcoin node. And much more if we talk about a pruned node, configured as the user wishes.
In fact, we at Etherscan are reminded that the problem continues to grow, as the network currently requires more than 900GB of space to store network history. The graph below shows us how fast that requirement grows.
The situation becomes more serious if we take into account that Ethereum 2.0 will increase the pressure on this requirement, especially with the arrival of Shards Chain, which not only parallelize the operation of Ethereum, but also trigger the need for storage and network resources to its operation.
The Ethereum Core Dev knows about the problem. The first attempt at a solution was the «pruning node», which occurred in GETH, but it was a temporary solution, which has practically remained a futile attempt. Options such as “snap” synchronization, capable of providing decentralized validation options, are already insufficient to control this measure.
The problem dates back to 2017 and although there has been a lot of discussion about it, the solutions do not help much.
The need for Ethereum resources does not stop
Now, what relationship does all this have with the problem of the centralization of Ethereum nodes? Well, first of all, just check your PC and see how much space you have available on your hard drive. Surely, it is a long way from the need for Ethereum at the moment. 2 TB of disk is not a small thing and not all personal computers reach that capacity. If we add to this the problem that this need grows over time, it is understandable that specialized solutions are required to have a full Ethereum node. The cheapest is the cloud.
First of all, the cloud simplifies the installation of the nodes. Amazon, for example, allows you to install Ethereum nodes in a couple of clicks, also having high bandwidth to perform synchronization in a very short time and disk space can be contracted for several years at a fairly low price. Installing your own computer, buying hard drives, making a RAID array on them and having a high-speed Internet connection for node synchronization is much more expensive.
Centralization is cheap
Centralization is cheap, but it is at the cost of your freedom. Decentralization is expensive, but it offers you freedom. In Bitcoin, for example, the metrics of the network in Bitnodes allow us to see that its operators are committed to the decentralization of the network in the broadest sense. In fact, almost half of the Bitcoin nodes are on the Tor network, protecting the privacy and identity of its users.
Against this, Ethereum’s need for resources seems endless, leading to problems like the ones described above. It would not be surprising if in the next two years residential nodes would fall by 15%, due to the need for network resources and how expensive these resources are.
Likewise, Ethereum lacks simple solutions designed for users to have their own functional nodes and that are easily managed. Options like DAPPNode help a lot with the task of installing an Ethereum node, but they do not solve the problem of the need for resources.
A more troubled future
If this course is followed, the future of Ethereum will be more centralized. Shards chains will require two things. On the one hand, high-speed networks to propagate blocks quickly and with low latency, in order to prevent third parties from attacking them. And, on the other hand, the need for each shard chain and its associated nodes to store their own history together with greater storage pressure on the mainchain, due to their increase. It’s basically a snowball effect, requiring certain points to be worked on before moving forward.
At the moment, the problems are multiplying and Ethereum has no solutions to deal with them. Shards chains will make the need for high-speed, low-latency connections a must. Without these connections, the structures of these networks may be vulnerable to certain attack vectors (such as timings attacks). The connections in the world may seem fast, but the reality is that the average world connection is about 31 Mbps. An insufficient figure using the shards. In Polkadot , which uses a sharding system, they have already realized the problem and are focused on making this system very efficient. At the moment, they have stopped at a total of 100 shards, while in Ethereum they dream of 1,000 and more.
Away from users and close to corporations
The decentralized solution to the problem is not simple, but the centralized one is: make the shards have a double structure, where there are validators and storage nodes. The first with weight in the consensus and the second, simply, to store data. Interestingly, it is the same solution that has been applied to Ethereum 2.0, where the «Execution Layer» only records the data of the blockchain, while the «Consensus Layer» is the one that validates and has the real weight on what is accepted and not. is accepted on the blockchain. Basically, the solution lies in concentrating the validation nodes in high-speed spaces (Amazon, for example) and leaving the rest only to see what is happening, without participation and without the right to decide on the validation process. of network operations.
Everything described above shows us an Ethereum involved in a process in which little by little it is forced to close its doors to users, leaving only a few to participate in its network.