Something has to be the store of data. The double-entry model is a paradigm, not a serialisation, of accounting and financial data. I’ve written double-entry code using text files, key-value stores, and relational databases as the underlying storage. They all work, at varying levels of scale and manageability.