What is decentralisation? | Blockchain Basics Part 3

Alexandra Matthews
Crypto Alexa

--

As we have moved the storage of data from physical ledgers into digital, there have been various advances in the available architectures of how we actually store data or ledgers.

Understanding the benefits of the distribution of blockchain data requires having the context of why prior data architectures were not adequate to create a blockchain.

The main data structures descriptions that you will usually find discussed online are centralised databases, decentralised databases and distributed databases. They normally come coupled with the diagram below and often contradictory definitions, which can be very confusing.

There are actually three types of decentralisation that you need to be aware of, in order to understand the difference between all of these data architectures.

Three Types of Decentralisation

Physical Decentralisation

The actual physical architecture of how the data is stored. If it is stored on more than one physical computer, this can be considered as physically decentralised. If however all data is stored on one computer, it is physically centralised.

Logical Decentralisation

Logical decentralisation refers to how the data is distributed on the nodes (multiple computers). If the data is replicated (so every copy is identical) it is considered logically centralised. If it is fragmented, so different parts of the database are stored on different nodes, it is logically decentralised.

Political Decentralisation

Political decentralisation refers to how control is decentralised. If one central entity is in control of the database, no matter the logical or physical decentralisation, then the database is politically centralised. If however control is distributed among the multiple nodes in the system then it is politically decentralised.

Centralised Architecture

As discussed, prior to computing, banks and registries used to maintain physical ledgers to keep records of ownership of value. These were centralised physical ledgers. The entire ledger was stored in one location and one entity was in charge of this ledger.

A centralised physical ledger is:

Physically centralised — There is a central location that stores one copy of the ledger.

Politically centralised — There is a central controlling entity

Logically centralised — As the data is not physically decentralised, the data cannot be fragmented.

After the introduction of computers, these physical ledgers moved to digital ledgers. These digital ledgers essentially had the same data architecture. They were physically, logically and politically centralised.

Distributed Architecture

Companies often decide to distribute their databases over multiple servers and locations. This can provide improved reliability as it removes the single point of failure in a centralised database and provides fault tolerance. It can provide overall higher computing power as well as cost reduction.

Sometimes the data in distributed systems is replicated and synchronized, but more often the database is fragmented over the multiple locations.

The key thing to remember with distributed architecture is there is still one central authority that governs the entire network. So while data that is distributed may be physically decentralised, it is still politically centralised as there is a central authority who owns the data and makes decisions regarding this data.

Apart from the often distributed architecture of large businesses operating over multiple locations, another example of distributed database architecture could be the first content delivery network, Napster. Napster was physically decentralised, logically decentralised but politically centralised (owned by a central authority).

General Decentralised Architecture

With the invention of the decentralised database, otherwise known as peer-to-peer technology, you now had the ability to have a distributed database, but without a central authority.

A decentralised network or peer-to-peer network is made up of individual computers (called nodes), which offer their computational resources to the network (storage capacity, data, processing power or network bandwidth). The rights and roles of each node in the network are equal and each node can sometimes be both a supplier and consumer of resources of the network. The more users/customers that participate as nodes within the system, the more powerful the system becomes.

The data within decentralised networks is:

  1. Physically Decentralised (Many copies of the data over multiple locations)
  2. Politically Decentralised (More than one controlling entity)

Prior to blockchain, networks existed which managed to achieve these two types of decentralisation, but as they were usually content networks, the data was decentralised (logically decentralised). An example would be BitTorrent or any file-sharing/content distribution network.

How is this different to blockchain?

In order to ensure all nodes within a blockchain network are aware of the latest transactions (in order to be able to properly ensure the validity of a transaction), blockchains need to be logically centralised, i.e. every node in the network had to maintain an up to date copy of the ledger. They have to have an up to date account of who owns which value. This means that a blockchain ledger cannot be logically decentralised.

Blockchain Architecture

Physically decentralised

No one person has control of the server, as if this was the case they would essentially be trusting a third party. In a blockchain, when we physically decentralise, we call every server holding data a node. Every node in this network holds a replica of data.

Politically centralised

We want to remove the trusted third party who would usually have control of the data. We do this by distributing control of the data to all nodes in the network.

Logically Centralised

The same data is stored on every node in the system. All changes to the ledger are propagated to all nodes within the network.

Vitalik Buterin noticed the confusion with decentralisation and suggested a classification network for identifying the type of data architecture you are dealing with. https://medium.com/@VitalikButerin/the-meaning-of-decentralization-a0c92b76a274

--

--