Scaling the compute, storage and communications infrastructure for applications is a major challenge in data centers. Dedicated resources are costly and may remain underutilized as applications typically run at only 20-50% of their capacities most of the time. Sharing resources may lead to congestions as multiple applications compete for the same resources at peak times.
When it comes to large quantities of data, e.g., in IoT scenarios, the transmission of all data into a single data center for processing may not the most viable option. A compromise between central and decentralized computing must be found.
The elastic scalability of IT infrastructures in accordance with (often ad-hoc) business requirements cannot be achieved with traditional data center approaches. Being able to quickly cover peak demands is essential in today's fast-moving business.
Another major challenge is the configuration and release management of a plethora of applications that are typically in use in larger organizations. Interdependencies need to be respected and considered in testing and upgrades. With a wider range of open-source components being integrated into business applications, esp. the topic of keeping up with latest releases and security patches is a major issue to be addressed.
Cloud services, edge computing and scalable storage networks provide the backbone of such elastic architectures. Modern front-ends need to be mobile, responsive and expressive. Everything from dashboarding, over interactive exploration and reporting, up to augmented and virtual reality is an option to provide actionable information wherever and whenever it is needed. Customer-facing websites may require a multiple of their usual resource if a campaign is run or high customer traffic occurs unexpectedly. Analytics platforms may run at low resource consumption for most of the time, until large-volume analytics are run - requiring multiples of compute, storage and communications resources. Manageable, elastic infrastructures offering a combination of cloud and edge computing give application owners this freedom. Private clouds may be federated with public clouds or IoT technology. A homogeneous application and configuration management view is facilitated by the consistent containerization of data repositories and business logic components.