Healing Enterprise Application Paper

Scaling enterprise applications at data layer

Introduction Scalability is an important factor for the organizations which do business on web that may suddenly become demanded. Scalability of an application is its ability to work properly even if its size or context is increased. It just not only means it should work well when its increased in size but also should take full advantage of it. Applications can be scaled at various levels. Be it at a data layer or API layer or the overall architecture of an application. Applications can be scaled up by increasing their processing power or distributing the processing at various nodes. This paper briefly discusses the various technologies for scaling an application at data layer briefly.

Oracle

Data virtualization enables to release the unused storage to virtual storage when there is a less load. Deploying a database instance in a virtual machine takes less time than traditional way of setting up a database. Database virtualization allows the administrators to add memory, CPU, disk or other instance with ease with very less downtime for users. Database virtualization makes the proper usage of the resources and allows to run the instances based on the demand, policy and constraints. Advantages include pooling unused resources, scalable compute, elastic compute, compute mobility and data mobility. The Oracle RAC conceptually also creates an abstraction layer of database by creating many instances of database without affecting the underlying

The I/O operation delays slows down the application and this effects the speed of the application when there are more I/O requests. The database should be able to handle increased requests with ease and good speed. Instead of directly relying on the database, we can use the in memory data nodes or In-Memory Data Grid(IMDG), a layer between the application and the database which provides high availibility, consistency and reliability while isolating the database from the application load and data persistence being handled in the background. The storage implementation and schedule are handled by IMDG and the application need not be aware of anything related to the data. The application can still ask for the synchronous database persistence and in memory persistence for more transient data. The application can achieve big boost in its performance without compromising flexibility. When a new server is introduced in the node then the node has to just connect with the IMDG and everything works as