The purpose of a node is to keep a copy of the network's database. Each node is an identical replica of the database, as long as it is up-to-date and fully synced. Anyone can create, or spin-up, a node. While they're not particularly resource-heavy in terms of computing power, they do require a substantial amount of hard drive space. Full copies of the network's database can be hundreds of gigabytes in size, and they only ever get bigger. Any users planning on hosting a node should be prepared to handle large amounts of data.
There are three different components on the Aion network: nodes, kernels, and the Aion Virtual Machine (AVM).
The reason that Aion has two kernels is for redundancy. If a huge failure or potential bug is found within one of the kernels, that set of kernels can be taken offline while the issue is fixed. In the meantime, the remaining set of kernels of the network can take the load and keep the network alive. For example, if there was a critical bug found within the Rust, all nodes running that kernel can be taken offline. The remaining nodes on the network that are running the Java kernel can pick up the slack left by the Rust kernels. Once the Rust kernel has been fixed, the nodes running a Rust kernel are turned back online and the Java kernel releases the slack it was holding.
Unity is a hybrid consensus algorithm in which miners and stakers work together to produce proof-of-work and proof-of-stake blocks to process the transactions on the OAN and to secure the OAN. Unity increases the security of the network and lowers the barrier of entry to participate in network security via staking.
Both full nodes and mining nodes are similar, in that they both have a full copy of the network database. They differ in the fact that only mining nodes process any data or requests on the network. Miners are responsible for taking requests and transactions from other users on the network and adding them into the network database. This process is called proposing a block as is very computationally difficult to do. This process is essential to the network, and the blockchain cannot work without miners providing this service. As such, miners are rewarded for their services in the form of crypto-currency.
Due to the difficulty of mining, the physical computers that run the mining nodes often contain multiple high-end graphics cards, as well as other expensive pieces of hardware. This makes them fairly expensive to build and maintain.
In the PoS subsystem of Unity, any coin holder of a system can run a full node in the blockchain network in order to produce PoS blocks; we refer to these coin holders as stakers.
In order to support a staker’s participation in PoS consensus, coins must be rendered immobile (i.e. they cannot be freely transferred to other users) via their submission into the staking system; these coins are referred to as stake. As long as these coins are held (on behalf of the owner) in the staking system, they contribute to the total influence of the staker in the PoS consensus (as discussed in the Unity paper and the Engineering Spec).
A validator on The OAN is considered to be an individual or entity that is running an operational node that is contributing to securing the network whether it is for PoW or PoS. There are different types of validators that encompass both algorithms, i.e. a PoW miner, a PoS staking pool operator, a PoS solo staker and so on.
As previously mentioned, nodes contain a full copy of the network database, and as such take up a substantial amount of disk space. This can be a problem, especially when a device will limited storage (such as a mobile phone or tablet) needs to access the database. Large storage requirements also pose an issue when trying to sync a node over a slow network.
One solution to this problem is to only store a small section of the database. Instead of having every row of data, a light node can contain a much smaller set, such as all the rowers entered into the database within the last month. This can drastically reduce the storage requirements, and increases the speed at which a device can find a particular piece of data. Light nodes come with the obvious drawback that they aren't able to access the entire historical data of the database, and as such are only applicable for certain use-cases.
Updated almost 4 years ago