Editor’s note - This post is the fourth in a four-part series on private cloud from Red Hat technology evangelist Gordon Haff. Read the earlier posts:
In previous posts in this series, we’ve covered private infrastructure-as-a-service (IaaS) clouds along with OpenStack specifically. We’ve also discussed container platforms from the operator’s perspective and how they can be used both to enable developers and simplify workload portability.
In their current form, these technologies are relatively new. They bring a lot of useful capabilities to IT operations. They also require management capabilities to evolve alongside. Hybrid cloud management needs functions like self-service access under policy-based control, metering and billing, intelligent workload placement, system image provisioning, capacity planning, governance, and lifecycle management—features that often go above and beyond what’s baked into the cloud infrastructure. At the same time, hybrid cloud management needs to fulfill its overarching goal of providing consistency across hybrid infrastructures.
The following aspects of hybrid cloud management such as those provided by Red Hat CloudForms are often of particular importance to enterprises.
Unified management and resource pooling across hybrid infrastructure
Even if a given organization isn’t using public cloud resources (yet), they are likely already hybrid in the sense of operating multiple infrastructure platforms, such as for virtualization; hybrid cloud management can help to unify these under a single management interface. Unified management can also give IT shops a consolidated view of geographically distributed virtualized resources for allocation, capacity planning, and chargeback purposes. Preferably, management should not require the use of agents, which can increase installation complexity and overhead. For an additional insight, view our webinar Unified IT Management and Policy Strategies.
Provide a public cloud experience for users under IT governance
That’s the one-sentence storyline underlying what organizations are trying to accomplish when they build a private cloud. They’re trying to meet the needs of users who are using public clouds because that’s where the users can get access to computing resources in minutes rather than the weeks common to traditional IT procurement. (And a private cloud project that doesn’t match the public cloud user experience might as well be shut down.) Cloud management can deliver that user experience while IT maintains oversight of access, quotas, and organizational standards. It’s about providing delegated administration and self-service for users while maintaining granular policy control for IT.
Orchestrates and monitors workloads across resource pools
Management involves having control and visibility over many aspects of workload deployment and ongoing operation throughout their complete lifecycle from deployment through retirement. This is needed, not only to make efficient use of the underlying infrastructure but to enable ongoing adherence to SLAs — which can be complex and multifaceted in the case of an enterprise software portfolio. And although relatively few organizations are implementing full chargeback pricing models at this point, an increasing number are starting to at least implement some form of showback as they shift toward a more services-centric view of IT application delivery.
Automated enforcement through policy
Consistent management is increasingly enabled by collecting data from running systems and taking automated actions based on policies that administrators establish. These automated actions may be as simple as raising an alert that notifies staff, or as robust as automatically remediating the problem. This can help meet compliance and governance requirements and eliminate many manual processes.
Provide the reliability and performance that users expect
There is a very strong, implicit assumption by users that their self-provisioned systems will be highly available and reliable, and will perform as required within the underlying cloud infrastructure. Providing only initial self-service provisioning without continuous monitoring and management of those systems, as well as the supporting infrastructure, will significantly constrain the service levels that your cloud can achieve. The management of underlying clouds must also be agile in order to handle fluctuations in user demand, to meet variable workload needs, and to continuously optimize resource allocation and utilization. Management requires robust back-end services such as service monitoring and dynamic resource allocation — not just a front-end portal.
Support the infrastructure that the operations and application development teams want to use
The best management tool isn’t much use unless it supports the right infrastructure. And that list gets longer all the time. At one time, managing just physical servers and virtual machines was a reasonable goal for a management product. Then private cloud and public cloud. Then more public clouds. Now containers as well. Furthermore, these resources may be layered on top of each other — for example, containers in OpenShift on top of OpenStack. A management platform should be able to map relationships between resources, provide capacity planning for future projects, and track the timelines of events for root cause analysis.
As you can see, the management of hybrid cloud presents a number of requirements for enterprises that are not easily addressed using traditional IT management systems. They’re just not a good fit with the dynamic nature of cloud environments and the speed of delivery required in those environments. Enterprises faced with these challenges should look for modern adaptable tools that specifically address hybrid cloud management needs. Because the one thing that the history of cloud evolution has taught us is that continued change is the one thing that we can count on.