The modern platform has three characteristics that distinguish it from the virtualized environments we’re all familiar with. First, the type of hardware required for a modern platform is vastly simplified. Infrastructure can shift from highly specialized, expensive servers to generic, repurposable and commodity hardware.
By Christopher Keller, Senior Technology Architect, CSC
Modern platform servers can be built from open data-center designs. Hardware is tied together with resilient, open source software that automatically protects and manages the environment. This bundle, connected to a network, becomes a defined quantity of storage, computing and memory that can be cut and redistributed into blocks of computing resource, billed on a usage basis.
This “hyper-converged” infrastructure starts with a set of basic building blocks of compute nodes that has a simple set of configuration options and is normally paired with software-defined networks, Infrastructure as a Service or Platform as a Service. The servers usually have four options to enable the most workloads possible. The most common building blocks are:
- A mixed node with a balance of memory, CPU and storage capacity
- A node with high-speed CPU and memory
- A node with high-capacity disks (6 TB SATA)
- A node with high transfer rate storage (Flash or SAS)
The adoption of this type of infrastructure is growing rapidly, and the uses are expanding regularly with VMware SDDC and Red Hat OpenStack. Replicated in blocks, the clusters of computing capacity can be tuned through predefined settings for different needs. For example, platform applications would have different needs than business applications. A modern platform also supports cross cloud-based storage solutions, including personal and group-based cloud storage.
Another distinction is that provisioning shifts from manual to automatic. Self-provisioning allows business units to request resources and build environments on demand, eliminating the need for IT to step in. The modern platform also makes it possible to support important advances in continuous application development and delivery. DevOps, which relies on tight integration between development and testing resources and the business, relies on the ability to rapidly move applications into different environments, depending on their current state.
Finally, workload management shifts from static to elastic. A major advance in platform design is the ability to move workloads between clusters of computing resources based on rules or on conditions that are detected when a workload is running. This allows the system to strike just the right balance, providing the amount of computing resources necessary to maintain a specified quality-of-service level without over committing resources.
For example, workloads running on an internal resource might be shifted in part or in whole to a public cloud infrastructure when demand spikes. After the spike is over, that extra computing capacity would be deleted. If a fault is detected in a particular node, awareness built into the system core can keep workloads running by shifting them automatically.
A comprehensive look at how digital can power forward the enterprise: csc.com/power-forward
Read more in this paper, Cloud Platform Enables Business Agility.
This paper is part of the Journey to the Digital Enterprise paper series.