How Hyperconverged Infrastructure Can Optimize Your Hybrid Cloud
Combining data center resources in a single appliance can help organizations overcome challenges such as IT complexity.
- by Tim Kridel
- Independent Analyst and Freelance Writer | March 29, 2019
When it comes to business agility, operational simplicity and cost-effectiveness, it’s tough to beat hyperconverged infrastructure. HCI combines computing, storage and network resources under a single management layer onto a single appliance. Managing these resources in one appliance can simplify data center infrastructures and help IT teams avoid integration headaches. Those benefits are why many CDW clients now have upward of 80 percent of their workloads in HCI environments versus 20 to 30 percent just a few years ago.
Hybrid cloud is popular for similar reasons. This architecture provides a broad, deep selection of computing, storage and services that are available in public and private clouds. IT teams can choose the mix that meets their business goals, budget and compliance requirements, and tweak that mix as those requirements change.
Many organizations are finding that they can achieve significant benefits by combining these technology trends. By using HCI for their hybrid clouds, organizations are better equipped to address a variety of IT challenges, such as distributed workforces leading to distributed data.
“Gartner estimates that by 2022, 50 percent of enterprise-generated data will be created and processed outside the traditional, centralized data center or cloud,” says Kapil Bakshi, a distinguished systems engineer with Cisco Systems. “Data centers must follow the data to branch, remote and edge locations. HCI should be part of wherever the data resides, be it the cloud data center or remote office branch deployment.”
Overcoming Complexity, Fragmentation and Skills Shortages
To maximize the benefits of HCI and hybrid cloud, organizations need a solid game plan.
“We often hear from our customers that their hybrid cloud migration was not planned,” says McLeod Glass, HPE vice president and general manager for software-defined and hyperconverged solutions. “They often struggle with a patchwork of tools rather than a carefully executed and integrated strategy.
“Some of the items that customers often overlook or underestimate when considering hyperconverged infrastructure for hybrid cloud are unforeseen time and costs, fragmented, complex management and lack of cloud-capable talent and skills.”
One way to tackle the latter two problems is by choosing a solution that uses a hypervisor that is already familiar to an organization’s IT staff.
“Also, make sure the solution provides templates and automated workflows for provisioning, orchestration and updates,” Glass says. “Integrate data services like deduplication, compression and backup in your hyperconverged stack for better efficiency and data security from a single management platform.”
Managing Multiple Providers
Some organizations have built out multicloud environments in which they use multiple cloud providers in a hybrid cloud model. This architecture creates additional HCI considerations.
“An organization must still have the same network agility across these cloud environments as they do on premises, and have an HCI platform that works with the same orchestration tools across these multiple clouds so management of the environment is tangible,” Bakshi says. “Developers need a seamless environment to develop and run cloud-native applications, across on-premises infrastructures and public clouds. They are looking for consistency, standardization and ease of migration to and from clouds.”
For example, the Cisco Container Platform (CCP) supports Microsoft Azure, Amazon Web Services and Google Cloud, and is based on open-source Kubernetes.
“This provides consistency, standardization and ease of migration to public clouds,” Bakshi says. “CCP runs on-premises and provides integration to multicloud with the same agility and orchestration tools across the HCI environment.”
AI for HCI
Artificial intelligence provides new opportunities for optimizing HCI environments and workloads.
“Hyperconvergence in general is a very good architecture for machine learning because this is the architecture where you look at compute and storage at the same time,” says Gil Haberman, Nutanix senior director for product marketing. “With our hypervisor, AHV, we’re able to look at the utilization requirements across different physical nodes. “
Those requirements come together with machine learning, which helps identify where to place each workload, as well as some of the data associated with storage.
“If you rarely use a workload — disaster recovery, for example — maybe utilizing some kind of public cloud extension would make sense because you benefit financially from sharing infrastructure,” Haberman says.
Protecting Data and the Bottom Line
Hybrid cloud and HCI give lines of business a lot of additional resources. This newfound freedom can result in unexpected bills if IT doesn’t have tools to enforce policies across the board.
“The most challenging area is achieving a standardized way to govern across multiple clouds,” says Gil Haberman, Nutanix senior director for product marketing. “How do I ensure that teams that provision, for example, in public cloud zones don’t overspend? You need to optimize the cost and drive accountability.”
Governance tools also help ensure that lines of business comply with company security policies and regulatory requirements such as HIPAA.
“Most companies have very robust compliance practices on premises because that’s where critical data always resides,” Haberman says. “But as we move to multiple clouds, sometimes users don’t maintain the same level of compliance.”
Numerous examples in the news illustrate the need to protect data no matter where it resides in a hybrid cloud architecture. Deployment of basic security measures goes a long way toward keeping cybercriminals at bay in most cases.
“It’s not a complicated hacking operation,” he says. “It’s folks provisioning databases in the public cloud with no password protection and no encryption. That goes to show the main issue is simply enforcing policies effectively.”
Another upfront consideration is the level of hybrid interaction. The simplest model is process integration: The public and private clouds are logically connected from a budgeting and management perspective, but they’re technologically separate in terms of control planes and other aspects. The other extreme is a completely unified platform.
“This ties together the logical processes, the control plane, the data plane, the network connectivity and, in many cases, some kind of data connectivity that gives you throughput and latency service-level agreements,” says Deepak Mohan, a research director in IDC's enterprise infrastructure practice.
The middle ground is a shared control plane that oversees provisioning, auditing, management, monitoring and tracking of resources across public and private cloud services.
“This seems to be where most enterprises have targeted,” Mohan says.