Welcome to Hard Fork Basics, a collection of tips, tricks, guides and tips for updating you in the world of cryptocurrency and blockchain.
Climbing the blockchain is becoming one of the great mysteries of technology. Does anyone know how to do it? Can it also be done? Some seem to think so, but obviously it will not be easy.
There are numerous "solutions" offered by various developers, researchers and academics. One of these solutions is known as "sharding". In this article Hard Fork is about to take a look at the hood and see what this blockchain sharding is.
Sharding is not new
The concept of sharding existed long before it became part of the lexicon of the blockchain industry. It is traditionally used in database management to ensure that the database works at optimal speed and efficiency.
Traditional sharding does this by separating a database into a series of smaller databases that remain connected, but they spread the workload on a large number of servers or nodes on that database network.
In general, none of these small "sharded" databases should have something that is replicated in another fragment to reduce duplicate data. This also serves to increase the efficiency and reliability of the database by ensuring that each entry is unique.
In a database where records are not interconnected and follow a fairly simple data structure, sharding is a relatively simple process. All system architects have to do is create a database structure that works safely and efficiently for its given use case.
This could include geographically identifying databases on servers in countries most relevant to archived data. This is fine when your database is centralized and you have complete control over it, so what happens when you split the blockchain?
Sharding the blockchain is, umm, complicated
One of the great strengths of blockchain is decentralization, so doing anything that could compromise it will make the system weaker, and likely susceptible to attack.
Since all the transactions stored in a blockchain are linked, it makes the sharding of this type of database even more complicated. But the principle remains the same, if we can cut the blockchain, we can increase its throughput and increase the number of transactions it can handle every second – which we are told is a good thing.
The way a blockchain can be sharated depends strongly on the underlying consent mechanism of the blockchain – by its nature a blockchain based on Proof of Workwork is very difficult to fragment. Because this would result in transaction validation without having access to the entire transaction history, new transactions must be validated without the history being known. Obviously, this creates a bit of trouble.
However, the cryptographer Peter Todd thinks it can be done. It would require another component to add to the operation of the blockchain, something called "proofchain". A proofchain actually maintains the historical integrity of the transactions, without having to refer to the entire database. It can be thought of as a control track, one that could be used to punish dishonest miners in case the need arises.
In the case of Proof-of-Stake blockchains, sharding is a little more attainable. In Proof-of-Stake there are dedicated nodes that deal with validating transactions. These nodes can only validate the relational transactions to the value of cryptocurrency that they are willing to bet for the opportunity. This encourages decentralization and prevents a node / staker from having too much energy.
However, because the nodes only need to validate part of the blockchain transactions, the database could be sharked and shared with each other, reducing processing and increasing overall network throughput.
It is likely that the downsizing of the blockchain will continue to be one of the most controversial topics in the industry. It is impossible to dramatically increase the throughput of true blockhain-based systems without altering the underlying structure. In this way, some of the main advantages of blockchain, such as decentralization, could be compromised.
Published 18 January 2019 at 14:11 UTC