Companies like Google, Facebook, and Amazon are online businesses that depend entirely on their digital infrastructure. Despite the struggles most businesses have in managing and leveraging IT effectively, those digital companies are able to manage their gigantic technical structure dynamically and responsively. Replicating that web-scale architecture and agility in an ordinary enterprise is becoming a goal for many other companies.
Hardware to Support Web-scale Computing
A web-scale architecture is reliant on infrastructure that scales out easily; rather than increasing the capacity of individual nodes, the infrastructure relies on many nodes and utilizes resources from wherever they're available. The infrastructure needs to have scalable capacity to meet compute, storage, and network demands.
The true web-scale companies—Amazon, Facebook, Google—custom build their data centers to meet those demands. They even custom-build hardware specific to the requirements of their workloads.
For ordinary enterprises, custom-built hardware isn't a likely solution. Instead, companies turn to hyperconverged platforms that make it easy to scale commodity hardware. The scalable design means businesses don't need to build excess capacity into their data centers but can scale affordably on demand. Through replication and virtualization, the services are highly reliable. Because compute, storage, and networking are virtualized, the infrastructure can be repurposed when requirements change. This virtualization also makes operational management simpler.
Software to Support Web-scale Computing
It's that operational simplicity that's another key to the success of web-scale in the enterprise and of the true web-scale companies. Speed and ease in deploying hardware is only half of what makes those companies successful. It's the ease and speed with which they're able to deploy software that lets them leverage that hardware effectively.
That requires adopting an agile mind-set and a DevOps approach to application development and support. By having a development process that brings developers closer to the end users, they're better able to understand requirements and more quickly implement applications that meet the business's needs. By bringing the operations team that will deploy and support the application closer to the development team, the applications are built with production support requirements in mind. The software development process becomes nimble from requirements through test and roll-out, and is able to more quickly respond to changes and problems along the way.
The software that's developed, too, needs to be built for agility. Because the environment changes so rapidly, software can't make assumptions about its environment. It needs to be able to discover and adapt to changes. The applications need to be built around APIs and standardized methods of accessing services and resources, without depending on the services and resources coming from any specific source.
Culture to Support Web-scale Computing
These new approaches require cultural changes as much as technology. Incorporating hyperconverged platforms means adding a commodity to your data center; incorporating agility and DevOps requires a change in mindset. It requires management support of a culture that works in an exploratory way, making frequent small changes and rapidly responding to adjust to any problems that result.
For a company to succeed at web-scale, management needs to accept that, just as a hyperconverged platform brings compute and storage together in a new model, web-scale brings together IT development, support, and operations into a new model. It's only by adopting the new IT model that a hyperconverged platform can help a company achieve web-scale success.