Application awareness is a concept whereby infrastructures, such as storage systems, are tailored to the programs that they support in an effort to deliver more performance, usable capacity or efficiency, and to make them easier to manage. Through software, systems can be made “smarter” and able to configure themselves to the variety of applications they run. In this Technical Insight we will discuss the potential benefits of making hyperconverged infrastructures (HCIs) more application-aware and review how HCI vendors are implementing this with a strategy of “VM-centric” operation.
More information on the vendors mentioned in this report, as well as others in the HCI market space, is available in Evaluator Group Research materials. These include the Hyperconverged Evaluation Guide, Product Analyses and Product Briefs, which include the Eval(u)Scale for Hyperconverged Infrastructure appliances.
Before VMware popularized server virtualization, each application usually ran on a separate server allowing things like storage devices and RAID level or backup and recovery parameters to be tailored to the application. The consolidation that hypervisors brought forth eliminated some of the costs associated with buying multiple servers enabling multiple applications to share the same host. However, this reduced the ability to specialize the storage for each application and created a one-size-fits-all paradigm. While hypervisors allowed the amount of CPU and memory resources to be customized for each virtual machine, all the VMs on a host used the same basic storage configuration.
HCIs and Consolidation
Hyperconverged infrastructures cluster similar (or identical) nodes, pooling resources to support larger data sets and more VMs. They can also support more kinds of applications, potentially increasing the level of consolidation and further “homogenizing” the infrastructure. This makes HCIs easier to set up and operate but can also make them less efficient with resources and potentially less effective at providing the best compute/storage environments for those varied applications.
Most HCI solutions configure storage properties at a cluster or datastore level. This ends up forcing multiple applications to use essentially the same resource configurations or data services. While not a one-size-fits-all situation, it could be called “a few sizes fits all”.
For HCI clusters that are implemented to support a specific application this isn’t an issue, since a few storage configurations would probably be adequate. But for hyperconverged systems that are set up in a more general purpose configuration and used to consolidate infrastructure (and many more applications), it can be a problem.
Download now to read this full report!