A number of solutions for this issue have been proposed but Duffield, commenting on Medium, feels that Dash offers a unique and total solution to the issue. Not only that, he intends to keep everything on-chain.

Why Scaling Is Necessary

Today, cryptocurrency transactions still represent only a very small fraction of all financial transactions worldwide. In order for true mass adoption to occur, cryptocurrency networks will need to support hundreds or thousands of times more transactions per second or TPS, than they do now.Reviewing the top three cryptocurrencies today — bitcoin, litecoin, and ethereum — they currently support approximately 7, 28, and 20 TPS, respectively. By comparison, PayPal processes an average of 193 TPS and can easily handle twice that. Payment network behemoth VISA currently averages at 2000 TPS, and according to official VISA sources, can support up to 24,000 TPS.Scaling is also incredibly important as it can have a big influence network fees. Over the last few months, bitcoin transaction fees have skyrocketed. While smaller, more nimble currencies like litecoin and ethereum have kept lower fees, they are definitely on an upward trajectory as their usage continues to grow.As more and more transactions are put onto a network strained, those trying to send funds will need to add increasingly higher fees to ensure their transactions go through.Some proposed solutions to the scalability issue, at least in regards to bitcoin, have included increasing the block size, or by not limiting block size at all.

On-Chain Versus Off-Chain Scaling

Duffield’s post is all about how Dash intends to enable what’s known as on-chain scaling. Before going into what that means in this case, let’s go over what it means to be on or off-chain.When we think of a normal cryptocurrency transaction, we are thinking of one that is “on-chain”. A person sends funds to an address, the transaction is added to the blockchain, confirmed, and then the transaction is complete. The advantages of on-chain transactions are that they are secure, transparent, and simple. The disadvantage of on-chain transactions is that they consume potentially scarce network resources, and may come with larger transaction fees.By comparison, an “off-chain” transactions occur first outside of the blockchain, and are then settled there later. An example of this would be several hundred small transactions happening on a separate off-chain network so as not to incur fees, then having the results grouped together and settled on-chain as one or two larger transactions.This concept of combining many transactions together before putting them on the blockchain is an attempt to reduce fees and, in some cases, increase speeds.The Ethereum-based Plasma network that is currently in development is one possible solution for off-chain scalability. Recent reports, based on statements made by creator Joseph Poon, imply it can theoretically support billions of transactions per second.This has yet to be seen, however, as the Plasma network has not officially begun operation yet.

Duffield’s Solution for Dash

In his piece titled How To Enable On-Chain Scaling, Duffield explains how Dash uses what he calls a two-tier network. This is in reference to how Dash employs both X11 mining, and what he calls “incentivized masternodes”. Additionally, the paper outlines how Dash intends to stay ahead of the demand curve, in order to support over 50 million users per month.In order to stay ahead of demand, the group plans to increase Dash’s on-chain block size to 2MB by the end of 2017. This will be followed by an increase to 5MB, 15MB, and an incredible 45MB through the use of “ultra-large block acceleration” or ULBA hardware designed specifically for Dash.In order to support this network, the group is working on developing a two-tiered network system. The two tiers are essentially masternodes and miners.

The Future of Dash Masternodes

Setting up a Dash masternode, which is like a special server that processes Dash transactions quickly in exchange for a portion of the block reward, currently requires an always-online server and 1000 DASH (approx. $320,000 USD at press time) as collateral.According to the upgrade plan, 20MB blocks can be achieved with standard hardware. By phase two of the plan, masternodes will need to be “customer computers with multi-core high-end processors, colocated in server farms for fast propagation and ultra-low latency.” In other words, not just the old Dell XPS sitting in your mom’s basement.When phase three is implemented, custom hardware will be needed in order for the network to be able to support the volume of transactions the group is planning for. The custom hardware will, according to the plan, be in the form of GPU based CUDA processors. The intent is that said CUDA processors can process hundreds of transactions in parallel.Phase four, the current final phase of the upgrade plan, outlines the deployment of PCI-EX peripheral based ASICs. Such devices could also be designed to operate over USB connections, which could potentially allow for hundreds of such devices to be connected and running on a single server.

Dash Mining in a Phase Four World

The next section of Duffield’s post outlines how increasing the speed and network throughput of masternodes in such a way might leave ordinary Dash miners in the dust. If such an event were to happen, those miners would lose their economic incentives to mine Dash, thus harming the network.Duffield’s solution is to move the entire mining process into the masternode network itself. Duffield refers to this as “collateralized mining”.Essentially, Dash mining will become impossible unless you are running a masternode, complete with the same DASH requirements. He claims that such a move will completely remove the risk of a 51 percent attack, as all miners must essentially be running a masternode. This is will also eventually remove all pooled mining.