Component Bursting

Overview

This approach de-couple a discrete functional subset of an existing application and enable it for cloud. The functional subset module must be easily packaged and may be replicated many times. Typically the subset is very read intensive with infrequent updates back to a central database. Often the component that is _bursting_has limited or no requirement for data persistence – a stateless object, or where data write-back is to a shared database via APIs.

This is great way to support on-demand scaling of critical high volume consumer activities such as Search, Quoting, Catalogs etc

Component Bursting

Use

A typical architecture might be to periodically replicate a relatively static dataset (e.g. a product catalogue) and then that dataset is cached for a defined time-period before being refreshed. That dataset (as it is static) can be replicated as much as needed to build the capacity required. If a transaction is needed (e.g. a product order taken) then the update transaction writes back to a central shared database. This means that the read-only capability can then be scaled independently of the write-back. This wrks for use cases such as a holiday search because the search capacity requirement can be 100x or 1000x the capacity required to make a booking.

Use Cases