A Hybrid Cloud is an organization’s core digital infrastructure that consists of multiple compute centers. These compute centers can consist of, classic brick-and-mortar physical data centers, collocated infrastructure, public clouds, and edge compute facilities (sometimes called micro data centers). Early in the adoption of the cloud concept a Hybrid Cloud merely referred to a mix of public and private cloud infrastructure. As cloud architecture and business models matured and became butter understood, it became clear that the Hybrid Cloud was the infrastructure that enabled application (workload) mobility to be positioned in the most optimal location to meet cost, user performance (SLA), risk and security requirements.
How is the Hybrid Cloud different from Public and Private clouds?
To start with, the basic premise of a cloud is to provide users some level of self-service in accessing services to deploy and manage applications and their associated workloads. Primary benefits for the user are application mobility and elasticity without dependency on physical resources to manipulate the infrastructure. Also, an organization hopes to reduce computing costs by improving efficiencies through better serve density and reduced human resources to manage moves, adds, changes, and deletes.
In some ways the term “Hybrid Cloud” is a misnomer, and actually better served by the term “Hybrid Digital Infrastructure”, but for now we can define Hybrid, Private, and Public clouds as is. A private cloud is simply a traditional physical data center of collection of data centers with an orchestration layer on top of it that allows for self-service. A private cloud is owned independently by a single organization on a private network. The orchestration layer in addition to user focused functionality acts as a traffic cop to control access levels and ensures isolation between accounts (users). A public cloud isn’t much different than a private cloud except for the network is public, and the infrastructure is owned by a third party. A Hybrid Cloud is an amalgamation of multiple compute centers and architectures with some kind of orchestration layer that focuses on workload optimization and the associated need for mobility. A Hybrid Cloud does not limit or change the functionality or specific management of the independent architectures beyond insight of the individual and collective performance of the architecture.
Why do I need a Hybrid Cloud?
The simple answer is to optimize application workload placement to achieve the best balance between cost and performance. The Hybrid Cloud gives you the ability to run workloads at the right time, from the right platform, at the right location, for the right price.
While most Hybrid Clouds today did not start out as a planned architecture, their very existence proves the need. Not all applications are created equal and have very different architecture needs, and thus the compute sprawl evolved. Monolithic architectures cannot accommodate performance, cost management, just-in-time resources, and security needed across an organization’s vast compute requirements. Many organizations have lots of money in real estate and the physical infrastructure of traditional data centers and to let that investment fallow is not responsible.
Organizations that have begun to look seriously in architecting a Hybrid Cloud (Hybrid Digital Infrastructure) are realizing there is about 20% of their application that need to remain centralized. Application that are remaining in the data center are those that are compute intense, manage sensitive data, or legacy systems that are not suitable for movement to the public cloud.
Colocation facilities and mini-data centers provide organizations the ability to move compute power closer to the user. Typical application are for high-speed retail transactions or IoT intensive operations perhaps for manufacturing automation and self-driving vehicles.
The public cloud exploded due in part to dev-ops teams needing environments set up quickly. Then since the application sandbox was in the cloud it made it easy to migrate the application as well. The just-in-time self-service made it easy for both developers and business teams to get application up and running without the historical IT overhead. As companies rethinking the “All Cloud” initiative they do realize that application requiring highly elastic architecture, tier 2 application, and born-in-the-cloud apps are ideal for the public cloud.
What are the components of a Hybrid Cloud?
Beyond the physical infrastructure management software makes the Hybrid Cloud. Gartner defines the management architecture of a Hybrid Cloud as Hybrid Digital Infrastructure Management – HDIM (which is a better description than Hybrid Cloud). The fundamental components of the management architecture of HDIM today require multiple vendor’s applications, but Data Center Infrastructure Management software – DCIM is the foundation which to start building from.
Other HDIM software should include IT Service Management, Automation and Operations, Cloud Management, and analytics tools. The goal of the management platform is to understand the performance and cost structure of every workload to best place it for its current state SLA requirements.
Crucial to an effective HDIM architecture is continuous discovery, machine learning and AI, and the ability to integrate communication between business systems through a single source of truth. This allows the brokering of data to broker data that allow individual of teams to make specific or collective decisions from the same, current data.
How did Hybrid Clouds come to be?
Most organizations today did not set out to build a true Hybrid Cloud, they developed organically. Independent groups within an organization set up specific compute infrastructures to meet application and user needs. Mergers and acquisitions leave organizations with multiple and sometimes incompatible systems and infrastructures. Cloud first initiatives drove many organizations to push applications into the cloud, at any cost. After some time, IT teams realized that certain infrastructures work best for certain applications driven by user experience requirements, cost per workload management, the need for elasticity, risk mitigation, and security concerns. Realizing the best digital architecture requires multiple infrastructure types, the need to manage the various parts as a single entity, a management architecture needs to be put in place. Hybrid Digital Infrastructure Management has become the new discipline and management architecture for today’s Hybrid Cloud.
How do you build a Hybrid Cloud?
Since the compute infrastructure is already in place, to build a hybrid cloud architecture out of it is really about the management of it. To build a Hybrid Digital Infrastructure management (HDIM) practice you need to start by combining management tools from IT service management (ITSM), data center infrastructure management (DCIM), cloud management platform (CMP), and other disciplines. The four key objectives are:
Understand the outcomes of various implementation strategies leveraging machine learning and AI to determine the optimal placement for a workload needs:
- Performance – latency, traffic, bandwidth, saturation, etc.
- Availability – resiliency and recovery strategies
- Orchestration – deploying within IT guidelines
- Standardization – of offerings
- Continuous evaluation of integration and delivery configurations
- Validation and audit of request fulfillments, patches, and backups
- Sandboxes for continuous improvement development
- Fault remediation and root cause analysis facilitated by machine learning
- Visual mapping of network, infrastructure, and workload dependencies for AIOps
- Identification of threats, vulnerabilities, and risks to value
- Current state visibility of infrastructure from IT facilities and endpoints to virtual, IoT, edge, and cloud environments
- Verify compliance with software licensing and regulatory requirements
- Regulation (GDPR, SOX, HIPAA, PCI)
- Application utilization data
- Network and infrastructure performance impact on service levels
- Audit log data for insights and threats
To select the vendor solutions and tools you need to implement the HDIM architecture you need to ensure they provide:
- Open APIs to allow information transfer and simpler integration between tools
- Common formats for messaging, alerts, KPIs, etc.
- Integrated dependency maps between the application stack layers
- Automated discovery and inventory of all infrastructure
- Mapping of workflows, network paths, and performance