EMA sees cloud computing as a fundamental game-changer, which will likely have an impact equivalent to that of the Internet on ISVs and IT organizations alike.
When considering the cloud from the perspective of the application, it becomes clear that cloud-based applications must be supported with the same rigor as those hosted internally. Regardless of the delivery mechanism, IT organizations still have a responsibility to end-user customers for application quality, availability, and performance.
Assessing this responsibility in context to the five characteristics outlined below – the main factors EMA has identified that distinguish cloud-based deployments – can help clarify requirements for cloud-related enterprise management toolsets.
1. Convenient, On-Demand Access
Convenient, on-demand access implies application availability, basically an assurance that the cloud-based service is accessible when the customer needs it. It also implies that applications are designed with easy access in mind. For applications developed in-house to run on public or private cloud, ensuring on-demand access requires toolsets that support application quality throughout the application lifecycle. It also requires tools capable of monitoring and managing application performance and availability from the user perspective.
2. Shared Resource Pool
For either public or private clouds to deliver economies of scale, a shared resource pool is critical. The idea is that different customers will have different resource requirements at different times. A shared resource pool — particularly when combined with the next capability (rapid provisioning and release) — means that customers experiencing peak resource requirements have access to pooled resources because other customers are running at non-peak levels.
This point and the next presuppose an ability to precisely assess utilization trends against capacity. Developing this capability requires management solutions capable of assessing and tracking the capacity of physical and virtual resources in context with one another, and with real-time and trend-based utilization.
3. Rapid Provisioning and Release of Resources
Rapid provisioning and release of resources are critical capabilities, and methodologies for release are equally important as — if not more important than — for acquisition. Rapid provisioning presupposes the use of products capable of provisioning applications on demand based on preexisting models and templates. Products (and processes) that support and enforce the governance of provisioning and release functions are also critical.
4. Minimal Service Provider Interaction
The average corporate user requesting access to an internally-hosted platform or service interacts one or more times with IT. However, when the same user seeks access to a cloud service, he or she expects to access the service directly, without an intermediary contact.
This ease of access has been one of the factors contributing to the cloud wildfire, as departmental credit cards (versus budgeted line items) have become the new “coin of the realm” for public cloud services. However, this de-facto de-centralization is also raising governance, control, and security issues, which again drives requirements for new kinds of cloud-related service catalog and service portfolio management solutions.
5. Minimal Management Effort
Public and private clouds, of course, require different levels of management effort, but the responsibilities of IT organizations are similar in both cases.
For public clouds, management effort, though “minimal,” is still a consideration. The base services delivered by public cloud providers are managed by the vendor. This is true whether the product is Infrastructure as a Service (IaaS), Platform as a Service (PaaS), or Software as a Service (SaaS). The business customer still has a responsibility to choose the vendor that most closely matches business requirements and to monitor services to make sure they are delivered at contracted levels.
Private cloud requires significant management effort, typically by internal IT. It also requires toolsets that support all aspects of the application lifecycle from “cradle to grave.” In addition, since the majority of private clouds are built using virtualization, supporting them also requires toolsets that “understand” and have visibility to virtual environments.