With CumuloNimbo, companies will drastically reduce the complexity of the applications they develop, as well as related personnel costs for developing and maintaining applications. This is thanks to the transparent scalability for transactional data management provided by CumuloNimbo.
The project ended in September 2013. Project coordinator UPM, together with other project partners, is now setting up a start-up to commercialise the results:
Transparent ultra-scalable transactional processing for cloud data management – Current cloud computing infrastructure is not able to scale with transactional processing and so loses transactional properties. Companies incur high personnel costs due to the higher complexity of developing cloud applications without transactional support. Using CumuloNimbo’s results, companies developing cloud applications will scale their transactional data management in a totally transparent way, resulting in significant costcutting for cloud application development.
Scalable SQL processing – Many companies developing cloud applications have opted for NoSQL data stores to circumvent the lack of scalability of regular SQL databases. However, this approach increases the costs of cloud application development due to the lower level interfaces provided by NoSQL data stores and their lack of powerful query capabilities. Using the CumuloNimbo technology, cloud application development companies will be able to perform full and powerful SQL queries and scale their workloads. This will result in a significant drop in development time and costs for cloud applications.
Highly concurrent NoSQL data store – Current NoSQL data stores such as HBase (used by Facebook, Yahoo, etc.) are optimised for offline batch processing. When they are used for online workloads they exhibit low concurrency, which forces the use of many more nodes than needed for a given load. The new NoSQL data store created in CumuloNimbo has been designed for online workloads, so companies are able to reduce the number of nodes used and save computational resources in the pay-per-use public cloud model. This cuts costs for cloud resources.
(Article from net-cloud future magazine (2013) - for complete magazine click here)