What can We Learned from the Crash of Infura API
Last year, when the author and others discussed whether the current blockchain is decentralized or not, we mentioned a point of view — -now in addition to Bitcoin and ETH, most of the tokens are centralized. Decentralization may seem to be a fraud in most projects, and most of them are in fact centralized.
Although the point of view is sharp, it has aroused opposition from others. However, this is the case, and people just don’t want to look at it directly. Recently, the API outage of Infura, the infrastructure service provider of Ethereum and IPFS, is actually a typical example that blockchain infrastructure services are still too centralized.
Is it really decentralized when there are a lot of nodes?
Now, as we all know, blockchain projects and digital currency is divided into two categories, one is the PoW mining mode, another is the PoS. Many people think that the decentralization of blockchain is based on the number of nodes, therefore, many projects try their best to add nodes, but in fact, all nodes run in a cloud service manufacturer, and the cloud manufacturer is the final unified manager, as well as centralized node management.
Later, as PoS increasingly become popular, and derived a lot of improvement based on PoS, such as DPoS, etc., decentralization becomes even weaker. If, take XMR for example, “one CPU one nodes” is well accepted, then PoS is basically unqualified. Thus, they started to make a lot of new tricks, but the PoS is still the same, in fact, the centralization is more serious.
For a typical project, such as EOS, instead of giving a detailed description, let’s simply talk about this. For any DPoS project, is it necessary to have a node, or is it necessary to staking tokens? Then, what is behind it? Are the servers really built by themselves? In fact, most of the PoS nodes we talk about are basically built on cloud servers.
As for common domestic cloud servers, there are such as BaiduCloud, TencentCloud, AliyunCloud, and as for abroad, there are such as Amazon, Microsoft cloud services. As for now, of course, the most popular one is Amazon. Besides, once the author also jokingly said that Amazon is the largest blockchain enterprise because most of the project nodes are hosted by Amazon.
Imagine, if one day when an accident happens in one of Amazon’s computer rooms, such as a power outage or a network outage, could that cause some major projects to fail? Here I think this is almost certain.
After all, it is not realistic to buy your own server and then put it at home or in the company like mining machines, because you have to face all kinds of risks, such as sudden power outage and network outage. And in order to keep the operation safe, you have to equip your own UPS/ satellite network to prevent accidents. In addition, the physical security of the server is also very important. For example, it is necessary to guard against thieves and other things, in addition, if there is a natural disaster such as fire one day, it will cause harm. The current DPoS and other models are basically very difficult to solve these risks.
What may even more dangerous is the possibility that all nodes of one project are hosted on Amazon. If we think of Amazon’s cluster of servers as one big computer, then the hosted nodes are like a simulated Intranet, which is the most frightening thing of all, and this will likely happen in the future of Ethereum.
What we see today is a basic API problem, although the impact is not very great, if we look at it in a deeper way, there are still a lot of problems, which all kinds of project parties need to pay attention to in the future.
So what is really decentralization in the core sense? A typical, the author thinks is Bitcoin. When Bitcoin’s full node wallet synchronizes blocks, the nodes are essentially connected point-to-point. However, API services are limited to light wallets, and the real full node is running on its own. You don’t need to host nodes on Amazon or Alicloud, as long as you own a computer, then you can make it, which is very different from most projects today.
Of course, if there’s an API problem, it’s basically because of the light wallet issue. Ethereum has a hard time developing a full-node wallet, and it’s actually because of the volume of block data, which is much larger than the full node of Bitcoin. There was an attempt to synchronize all the Ethereum blocks, and it took a whole month, which is hard to imagine, and the Ethereum wallet we’re using now is really just an API service for the full node wallet.
So is there a good solution for the future? In fact, there is. Here you can notice the IPFS data saving function, which goes something like this:
First, IPFS or blockchain decentralized cloud storage saves the blockchain data from the various projects, and then through the API service, provided by the middle of the corresponding component, finally, achieving the goal of decentralizing the whole node of blockchain. Because cloud vendors like Amazon are so centralized, the problem is easily solved once they’re replaced by on-chain storage.
It’s simple in theory, of course, but at this point, it seems that no public project team has announced to do so, because the incentive layer of IPFS is still lacking, and it will take some time to make up for this. However, from this perspective, of course, IPFS may play a significant role in the underlying storage aspects of the blockchain and various node services in the future.
Article from Cncoin
Translated by Yang(Mengyan Finance)