5.3.2. Our Latency Approach

Due to the evolving nature of our project, there are two ways of how we tackle the problem of latency. In the beginning, our network structure consists of a data center, which is connected to numerous Virtual Private Servers (VPS) all over the world. This star-network uses high-speed connections to the data center. We already have invested in Dedicated Internet Access (DIA) technologies, which allows us to use custom routing paths that are 100% to 1000% faster than regular internet connections.

The VPS machines themselves are not interconnected, instead they are connected to different exchanges in the best way that is possible for the respective exchange. Some of them already have a presence in the NY4, which is an advantage, because they can directly be cross-connected.

Not every one of our VPS nodes is connected to every exchange. However, in order to guarantee a fail-safe system, we create redundancy by connecting at least two different nodes via independent VPS services to one exchange. If one VPS service fails, the second one will keep working, so there will be no downtime for this exchange in our system.

By minimizing the physical distance between the servers of the exchange and their according VPS, latency can be reduced by a factor of 100. This is why we selected VPS services that have their servers in close proximity to the servers of exchanges.

Illustration 12: If you are located in London and you try to get market data from four cryptocurrency exchanges - e.g. Kraken, Bittrex, bitFlyer and Bitfinex - the total latency would add up to 700ms on average.

Illustration 13: With the improved, VPS-based solution that Future Gold Arbitrage is using, this only amounts to 295ms. In this example, Latency is cut by more than half

Beginning with the third stage of Future Gold ARBITRAGE, the network structure will become more decentralized. The Future Gold wallet, which already was capable of acting as a node in the Future Gold blockchain by staking, will get an additional feature, which can turn the users personal computer into a node of our arbitrage trading network (see Wallet).

Nodes with a low latency to a given exchange are then preferred to obtain current market data from the exchange, while nodes with high computing power are used for calculations. Each owner of a node is rewarded according to what the node contributes to the whole system. This creates an incentive to run specialized nodes, which are actually useful to the network and thus profitable for the entire community.