Cloud computing replaces supercomputers with a networked group of servers. It lets scientists build massive models of weather patterns. Cloud computing is also used in medicine to model newly discovered viruses, map genomes and track pandemics. Engineers have used computer aided modeling to run 3D computational models through simulations long before they made prototypes to run through real world stress tests. Cloud computing is now being used to assemble thousands of individual CAD models into cars and planes for environmental testing and stress testing. A recent advancement in computer-aided modeling has been the ability to use cloud computing to compare different product designs before combining their elements to generate dozens of novel designs. Each design is compared against performance criteria like energy efficiency, size and speed. Only the models that meet all of these criteria are returned for the designer's review. The process of combining elements to create several different prototypes leaves engineers with only a few models to build and test.
Software providers started offering distributed computing and applications that ran on their computers. This was an alternative to users having to buy and install software. This allowed users to avoid installation problems. They would simply receive the next software version automatically. Software vendors saw this as a constant stream of revenue from subscribers. Virtualization techniques from web hosting service providers for virtual servers were adopted wholesale by data centers.
More companies are adopting a hybrid cloud computing model. This includes private clouds for sensitive data, and public clouds for other applications. A company may use a cloud-based version of PTC's Autodesk to create drawings, but save them on a product data management application on a private cloud to keep proprietary drawings secret. Cloud computing providers will need to evolve to develop public/private cloud interfaces.
Software applications with high input/output cannot be moved into a public cloud, but may be run on a private cloud. Legacy software applications don't translate well to the cloud model, but can be run on a virtual desktop off of an executable file on the cloud server. Applications with low latency don't work well on the cloud. Proprietary data shouldn't be stored on a public cloud, due to security concerns.