Home > Technology > How World moved away from Standalone Servers to Cloud IaaS ?

How World moved away from Standalone Servers to Cloud IaaS ?

June 11th, 2016

Birth Rise of Cloud ServicesBefore wide-spread server virtualization (mid 2000s), corporate IT departments would order a server, wait days or weeks for it to be delivered, install an operating system (usually Windows Server) and then run that server, in support of one application or one part of an application. A significant issue in this pre-virtualization world was maximizing server compute resource utilization by properly matching a server’s compute resources with the necessary computing power needed for an application. Many times, servers would only be needed for part of computing resource, as the hardware needed to be sized for “peak” load. This led to significant inefficiencies in hardware, where as little as 5% of a server’s resource was used by a single application.

Then came server virtualization. Virtualization “slices up” server compute resources to where multiple applications can use the resources of one server, drastically increasing utilization. Each “slice” of a server’s compute resources is called a virtual machine (VM). Due to the significant efficiency gains wrought by using hypervisors to virtualize / slice up servers into VMs, this software infrastructure technology went from nearly 0% penetration of the worldwide server install base to >80% penetration in 10-12 years.

Just as server virtualization penetration was inflecting, AWS was quietly launched in 2006. What many did not realize at the time was pubic cloud infrastructure as a service (IaaS) promised similar (or better) efficiency gains as server virtualization. Essentially, with one (or a few) vendors specializing in being the world’s datacenter, costs decrease and the pace of innovation within infrastructure software increases.

Advantages of Moving Corporate IT to Cloud – First, economies of scale bring down per unit costs (AWS’s datacenter is exponentially larger than any one enterprise’s datacenter). As public cloud players gain more scale, their per unit cost comes down and open source software was the choice of enterprises. Second, customers can choose from a vast range of VMs that vary widely in terms of their compute resources and thus can increase server resource utilization by much more effectively matching applications with an appropriate VM. Third, customers can rent VMs by the hour and thus use compute resources only when they need it (allowing other customers to use that resource), again boosting utilization. Finally, given AMZN is running all customer workloads in their data centers, they can pass advancement in operational efficiencies on to customers.

Categories: Technology Tags:
Comments are closed.