According to the 2019 Enterprise Cloud Index, “85% of enterprises continue to rank hybrid cloud as the ideal operating model.” That’s the ideal, but how do you run hybrid cloud, a combination of private cloud and public cloud native instances, efficiently? How do you effectively manage the different operating models, management consoles, and tool sets required to operate both environments, while keeping the costs of running public cloud components under control? The answer is using state of the art hyperconverged infrastructure as the core of hybrid cloud infrastructure.
We’ve Seen This Before
Organizations used to run their applications on scores, or even thousands, of individual servers with utilization rates typically averaging less than 10%, which meant a huge waste of server resources alongside the expense of associated rack space, power, and cooling. Server administrators had to manage, patch, and upgrade individual physical machines. Virtualization changed everything by abstracting the physical servers into virtual machines, thereby slashing costs through consolidating the number of physical hosts required, and eliminating individual server management.
Virtualization also led to an explosion in storage arrays, as organizations across the globe started buying SANs to run VMware vMotion and hiring dedicated storage administrators to operate them. These proprietary storage arrays, along with their dedicated storage networks, were expensive, complex to manage, and did not scale well. The SAN-based infrastructures commonly took weeks or months to order and deploy and required multiple vendors to provide support. Vendors later offered prepackaged “converged infrastructure” to ameliorate these challenges, but to little effect.
Abstracting the Storage Layer
Hyperconverged infrastructure (HCI) entered the scene in 2012 with an entirely different approach to storage. Rather than simply provide a “faster horse,” HCI abstracted storage into a software application running on commodity servers. This freed organizations from depending on proprietary SANs, storage networks, and storage administration specialists. HCI slashed the complexity of SAN-based infrastructure while enabling fast and easy scalability.
The HCI architecture took hold despite strong initial resistance from the datacenter incumbents; today nearly every leading storage manufacturer offers its own HCI solution, and organizations around the globe continue to move off their proprietary SANs. According to IDC, HCI sales are now at “51 percent of the total $3.9 billion converged system market on a global basis.”
We’re now well into the era of cloud computing, and organizations regularly turn to the public cloud for its ease of use, elasticity, and, for some use cases, better economics. Yet public cloud providers use “t-shirt size” instances of vCPU and memory, which causes VMs to often have a small amount of leftover, unused resources. When aggregated across hundreds or thousands of VMs, this adds up to a significant amount of waste. Factor in CPU oversubscription inefficiencies, storage tiering costs, and an inability to hibernate inactive instances – and public cloud is often far more expensive than an on-premises HCI environment.
HCI 2.0 uses hyperconverged infrastructure as the anchor technology running in bare-metal servers in public clouds. It optimizes the public cloud by sharing unused resources among VMs – enabling more realistic oversubscription, utilizing the NVMe capabilities of bare-metal servers for increased performance, and including “hibernate and resume” capabilities--that is, the ability to stop cloud instances (and the associated billing) when they are not being used and to bring them back up when needed.
Hybrid Cloud Optimization Beyond Cost
HCI 2.0 should, in addition to reducing public cloud costs, also enable a vastly improved hybrid cloud operating model. This means easy migration to, and management of, public cloud instances without requiring application refactoring or additional IT staff or skill sets. There should be one management plane spanning both on-premises and public cloud, and administrators should use the same tools across both platforms.
Finally, an optimized hybrid cloud should eliminate the risk that comes from being tied to a particular provider--in case it raises prices, experiences security issues, or has trouble meeting SLAs. While virtualization lets organizations move VMs among server hosts, HCI 2.0 allows you to move VMs easily between public and private clouds. This mobility not only frees the VMs from dependency on a provider’s proprietary APIs and tools, but it also opens up new opportunities, such as enhanced disaster recovery protection by replicating from one cloud to another.
Abstracting Public Cloud
Just as virtualization once abstracted storage, and transformed enterprise computing, hyperconverging clouds through hybrid cloud infrastructure--HCI 2.0--abstracts complexity by enabling seamless VM and app mobility between on-premises and public clouds. This innovation allows companies to quickly embrace a modern, stable, and agile architecture that empowers you to always use the right cloud at the right time.
For more information on Hybrid Cloud Infrastructure, visit Nutanix.com.