Overview

To ensure that applications are performing to an acceptable standard, automated testing is required. Several types of testing are useful, including load testing, capacity testing, volume testing, stress testing and soak testing. All of these are similar in design, though different in precise specifications.

• Load testing simulates a specified number of users, each performing some predetermined set of tasks, monitoring performance and verifying that the tasks are all resolved in a timely manner.
• Capacity testing is very similar to load testing, but rather than using a specified number of users, it seeks to determine the maximum number of users that can be accommodated before performance degrades.
• Volume testing is intended to test the amount of data being processed rather than the number of users that can be handled.
• Stress testing pushes the limits of capacity and volume testing, trying to determine the failure points of the software by simulating demand far in excess of what is expected.
• Soak testing simulates high volumes of users or data for extended periods of time, to find problems with handling high traffic for extended time periods.

AMM Summary

An AMM, or Automated Market Maker, is designed to allow users to trade between two or more commodities, without the need to find a specific trading partner. The basic principle is to establish a pool of each of the possible traded resources, then fulfill trades from the pool. The variant we have implemented is a Modified Wombat AMM.

The Wombat algorithm was designed to facilitate trading between stablecoins. Our modification allows the AMM to account for changing relative values, as one would see in currency exchanges. The particular insanitation we are using for testing purposes, has two commodities roughly corresponding to British Pounds and Euros. An initial pool of $100,000,000 worth of each is established, and this is what is traded.

The algorithm allows the commodities to be traded near market rate when the amounts in the pools are balanced. As trades are executed and the pools move away from their equilibrium ratio, the relatively scarcer commodity becomes increasingly more expensive, whirl the more common becomes cheaper.

Each trade is charged a small fee, currently set at 0.03%, with half of the fee being retained as profit, and half being distributed to the pools, allowing the liquidity providers to earn an income on their investment.

All of our performance tests consist of the creation of an AMM as described, followed by the execution of a large number of bi-directional trades. This is a reasonably accurate scenario for the actual usage of such an AMM.