The need to support trends like social, mobile, data, and cloud is undoubtedly a necessity across the business. Hybrid-cloud offers a solution for enterprise architects to manage these platforms.
The hybrid model unlocks the potential of these trends by enabling access to data and services across organisation boundaries including tapping into data and services provided by others, and allowing others to tap into your data and services.
Businesses continue to ask what the hybrid cloud end-state looks like and how to get there. There are many hybrid cloud models each offering different compromises. The reality is that organisations will have an amalgam of hybrid cloud models.
It is imperative when assessing hybrid cloud models, to determine what is meant by the term “hybrid?” NIST defines hybrid cloud as “a composition of two or more clouds that remain unique entities but are bound together”. Enterprise architects require a more precise definition for evaluating the benefits and tradeoffs encompassed broadly within hybrid. Indeed, there is a vast spectrum of hybrid models that can cover a combination of deployment and architecture models.
Definition of a hybrid deployment model: Here hybrid refers to a solution that spans across two or more deployment models (that include public, private, virtual private, community). Beyond multi-site, hybrid deployment means that different organisations host the deployment and are responsible for the underlying service.
In the case of legacy and private cloud, the different organisations exist as part of the same enterprise but will have very different policies and procedures for management and operations.
Examples of hybrid deployment models:
- Hybrid: Cloud-Bursting from a private VMware based data center to a public one
- Not Hybrid: Virtualized Environment, Active-Active Data Centers Operated by the Same Organization, Virtual Machines on Amazon EC2 (Elastic Compute Cloud)
The hybrid deployment model leverages a mix of services from other entities. As lower level architecture components standardise, third party providers will specialise in offering unique data and services including functionality, content, cost, SLAs. A hybrid deployment model allows the enterprise to take advantage of these capabilities for agility, cost, and scale while focusing its resources on maintaining key services and supporting legacy systems.
Definition of a hybrid service model:
In this instance hybrid refers to a solution implemented using two or more service models within the architecture stack (e.g., legacy, IaaS, SaaS, PaaS).
Example of hybrid service model:
- A scenario may be a near perfect fit for PaaS. One solution is to drop the implementation of the entire solution to a lower less-abstract IaaS. Or a “hybrid service model” approach implements the majority of the solution using PaaS, and only the parts that can’t be implemented using PaaS use the lower service model.
The hybrid service model combines a number of standard services across different architecture layers. The many layers of as-a-Service cloud, offers ready-built service models that fulfil out-of-the-box functionality within the architecture stack. The cloud provider defines the functionality and handles the implementation specifics. When selecting a service model for a solution, the goal is often to select the highest level of abstraction suitable for the solution. The hybrid service model offloads commodity details to the provider thus focusing the organisation resources on what truly matters.
Examples of the application of this definition:
- Example 1: A bare legacy to cloud scenario comprises of a legacy application (non-cloud-based and non-virtualized) on one end and a cloud solution on the other. This bare legacy to cloud case satisfies the hybrid services condition in terms of different levels of architecture components, and may satisfy the hybrid deployment condition when the cloud resources are operated by another organisation.
- Example 2: A hybrid integration scenario uses a cloud-based integration platform (iPaaS) to ease cross organisation access. This hybrid integration case satisfies both the hybrid deployment condition where the iPaaS is operated by a 3rd party and may satisfy the hybrid service condition interfacing with other components on the architecture stack.
Defining the hybrid cloud taxonomy presents a necessary first step upon which architects can explore the different degrees of hybrid, determining the pluses and minuses of each. Cases and evaluation criteria are different when exploring hybrid deployment or service models and a definition that distinguishes between them, allows each model to be considered in its own right. For instance, addressing hybrid-related issues like data segmentation, portability, and performance requires different solutions when implemented across operational entities (with hybrid deployment) as compared to architectural components (with hybrid service).
Under this definition, hybrid cloud is pervasive and most enterprises have already adopted some form or another. However, taking the next step involves understanding what it means to adopt hybrid cloud in the context of its impact on issues like security, performance, portability maintainability, and what that actually means in terms of being able to quickly access new data and services.
Posted by Niel Eyde, Manager at Accenture and Teresa Tung Manager, Accenture Technology Labs