Jim Starkey, database veteran (see box), started developing the architecture for the new database in 2008. "At that time, the variations of existing database systems were exhausted. For a high scalability new ways had to be taken ", the database veteran describes the initial situation. The idea was to develop a very elastic SQL database that was geographically distributed and easy to extend, and that needed to meet the design criteria of the ACID (Atomicity, Consistency, Isolation, and Durability) database management.
Design in three layers
Traditional relational databases scale up. This means that larger machines must be purchased for more users or better throughput. The currently desired scale-out architecture was previously only possible with non-SQL databases. Starkey developed a system that did not follow either the shared-disk or the shared-nothing principle, but instead created its database as a peer-to-peer solution. And he divided the system into three layers: a management layer, a transaction layer, and a storage layer.In particular, the separation of the transaction level from the storage level ensures that NuoDB scales so well. "Because the layers are separated, they can be independently scaled and treat errors independently of each other," says Barry Morris, co-founder and executive chairman of NuoDB, the advantages . Therefore, it is possible that transaction throughput can be increased without affecting where and how the data is stored. Conversely, the data may be stored in different locations without affecting the application. "That's why our database scales on-demand and allows the use of automation tools," explains Morris.
application scenarios
Today's business requires a global presence, so that around seven billion consumers worldwide can be served. The only thing that succeeds on the internet is who can come up with fast response times. Experts believe that the load times should not exceed 0.3 milliseconds. From 0.6 milliseconds, customers leave the application. Fast transactions are also required for competitive reasons. In the near future, transactions will increase sharply, especially in the case of mobile payment systems. In 2010, goods worth just under $ 3 billion were paid for by mobile payments, and in 2017 they will already be over $ 700 billion.
