Built to Scale — Blockchains and the Scalability Concern

The issue of blockchain scalability is probably the one that raises its head most often, alongside regulation. Depending on the method of consensus, block size (if applicable) and underlying technologies being used, blockchain speeds — transactions per second or ‘tps’ — have been reported from anywhere between 2tps and theorised millions of tps. Scalability is the single biggest selling point of alternatives to Bitcoin and Ethereum and every blockchain is working on methods to improve. When examining this issue, it is important to keep in mind the following:

  1. Some blockchains are currently more popular than others and therefore have been tested to the limit.
  2. Theoretical transaction speed is not the same as actual transaction speed.
  3. It is easy to make claims but harder to prove them in a real-world environment where unseen variables come into play.
  4. Some blockchains are focussed on one use — e.g. bitcoin for payments.
  5. There is almost certainly going to be a trade off in one area (security, decentralisation, computing power or other) when trying to dramatically improve scalability.

The particularly key aspect to review to the problem is the number of transactions per second that the platform can process, and the related confirmation time once a transaction has been sent that this leads to; although, also of note is the amount of space that it takes to store a full node of the blockchain: if too much data is required, it limits the number of people that can afford to run the nodes which will restrict the growth and therefore security of the network.

Unless the number of transactions per second that a network can handle is sufficient, the blockchain will not be able to be used for a number of solutions to which it could hold the key, such as micropayments. As blockchain technology merges with the Internet of Things, Big Data and AI, there is a need to process potentially hundreds of thousands and even millions of transactions per second as connected devices interact alongside humans using the network. Similarly, unless confirmation times are low enough, using blockchain solutions are not going to be acceptable for everyday use cases.

Without addressing these concerns, development of blockchain solutions will slow and the real world applications of the technology will not be realised. This has been seen at various points over the last year with questions raised around Bitcoin transaction fees and the slowing down of the Ethereum network.

Despite being the most well-known system and proven for decentralisation and security, Proof of Work may not be the best solution in the long run as the time and energy it takes to validate transactions are prohibitive of scaling without additional fixes being worked on (more on this later). Proof of Stake holds more promise as a base but there are other ‘Proof of’ systems being developed, including subtle variations of existing ones.

When looking at the number of transactions per second, blockchains are often compared to PayPal, which has around 200 tps, and Visa, which averages about 2500 tps but has a peak of 24,000. Compared with the original Proof of Work Bitcoin blockchain, with 2 to 7 transactions per second (although 16 tps have supposedly been recorded) and a block time of 10 minutes, traditional methods have the upper hand. Even with Bitcoin Cash increasing the block size to enable 60 tps it is still a long way off.

Ethereum also began using Proof of Work but has plans to move to Proof of Stake. Currently it manages to process between 15 and 25 transactions per second with a block time of 15 seconds. This is superior to the original Bitcoin blockchain in terms of scalability but still not usable for worldwide adoption. As one of the largest blockchains, Ethereum has a lot of public interest and has been very vocal on its need to scale and the methods it will implement in the near future to achieve this.

An alternative to Bitcoin, Litecoin facilitates payments between businesses and people with roughly 56 tps. Charlie Lee, creator of Litecoin, plans for the network to be used alongside the Bitcoin blockchain — even when proposed scalability solutions come into force on both networks. Dash is in a similar position at the moment but use ‘masternodes’ to scale the platform which will simply pour more power into running the validating nodes as the network scales to achieve theoretical millions of transactions per second.

Ripple is a payment focussed blockchain and has focussed on banks, exchanges, payment providers and the corporate world. When it first began in 2012 it ran at about 80 tps but at the end of last year was recording 1500 tps, with 4 second block times. As a solution that works with financial services, Ripple may be expected to scale well as investment should be readily available. However, Ripple is somewhat centralised in banks and payment institutions so adoption outside of financial services may not be so desirable.

Like Ethereum, Neo is focussed on smart contracts like Ethereum. It uses Proof of Stake and has a theoretical speed of 10,000 tps but practically has achieved 1,000 tps. Originally focussing itself in China, Neo is positioning itself as a main competitor to Ethereum and is poised to attract significant attention in the next couple of years.

Using a ‘Delegated’ Proof of Stake system, Eos is another competitor to Ethereum and claims to remove transaction fees for end-users and facilitate 20,000–100,000 transactions per second with the potential to scale to millions of tps; although, as it is still under development these numbers have not been tested in the same way that Bitcoin an Ethereum, for example, have.

NEM is another blockchain using a different system which is known as Proof of Importance. It is focussing on real-world use cases and has been shown in testing on a private chain called Minjin to achieve 3000–4000 tps.

Designed for use with the Internet of Things, IOTA uses a novel technology called ‘tangle’ which is designed to simultaneously solve scalability issues and eliminate transaction fees. Using tangle technology, IOTA transaction times increase as the number of users grow.

A mix between Bitcoin and Ethereum, Qtum’s mainnet processes about 60 to 70 transactions per second. Scalability for Qtum can occur via its Decentralised Governance Protocol if the network members decide, but the development team is also working on the Qtum x86 VM to improve efficiency an tps.

Stellar is becomingly increasingly popular and has a reported 1,000 transactions per second, with the potential to double this without much difficulty whilst using its Bitcoin-inspired Stellar Consensus Protocol. It is planning on implementing the same scalability solution as the Bitcoin blockchain.

Private blockchain Hyperledger Fabric does not tend to publish its transaction rate but there is reference to it achieving in excess of 3,500 tps and uses a Practical Byzantine Fault Tolerance System which has been described as scalable to tens of thousands of tps.

Zilliqa uses sharding technology to achieve thousands of tps. Its testnet has processed between 2,000 and 3,500 transactions per second. Like Hyperledger Fabric, it uses a Practical Byzantine Fault Tolerance System.

There is currently a huge range of blockchains available with various speeds and scalability prospects. Whether the theorised tps rate that some are claiming are valid remains to be seen but each blockchain is developing scaling methods.

Lightning networks are being investigated by blockchains for off-chain scaling whereby private channels can be opened between network participants and the result only brought on-chain at the conclusion. Bitcoin, Stellar and Ethereum (in Ethereum Raiden) are some examples of where this is being brought in. Tangle technology, as being used by IOTA, promises amazing potential scalability but is still early in development and needs time to prove itself. Zilliqa’s approach of sharding is another potential scalability method whereby the network would be split amongst different nodes which would then work in concert to verify transactions — increasing the processing power, and therefore scalability, available to the network. Plasma is yet another scalability solution which is also being investigated by Ethereum which creates a hierarchy of blockchains which can share the workload of the network and only require the ‘root blockchain’ to be updated occasionally.

Whichever blockchain you examine, scalability is a key focus and one that is drawing huge attention from the community to determine if blockchain solutions are viable in the real world on a large scale. Whilst there may be not solid answer yet, with so much research and development on the topic it is only a matter of time before a workable solution, or solutions, are implemented. Whether it is the lightning network, tangle, sharding, plasma or another yet-to-be-invented solution, 2018 seems like it will be the year that scalability will come.

— By Matthew Warner