The use of cloud-based platforms in the technology industry continues to evolve into more complex multicloud arrangements involving a mix of public and private cloud resources. Why? The business world now demands many best-of-breed cloud services to form the optimal solution. Cloud Management Platforms (CMP) are fast becoming the critical component for successfully leveraging multicloud environments.
What are Cloud Management Platforms?
A Cloud Management Platform is a suite of integrated tools that provide automated management of private and public clouds. CMPs facilitate the operation and build out of cloud services by eliminating the need for cloud silo specific interfaces and end user knowledge of cloud underpinnings. Key capabilities include:
- Self-service interfaces for:
- End user requests
- Self-service build, provision and launch of images
- Metering and billing of resource consumption
- Workload optimization and governance via:
- Role Based Access Control (RBAC)
Many vendors claim to provide full CMP functionality. In reality, however, only a few vendors can deliver a comprehensive function set.
Importance of Cloud Management Platforms
CMPs are critical to any medium to large organization that uses multiple cloud endpoints, and is hosting more than 5,000 virtual machines. A good CMP ensures that applications and workloads are running in approved clouds based on established service management, governance and compliance policies. Adhering to compliance requirements becomes even more important when data or applications fall under HIPPA, Sarbanes Oxley, FERPA or any other regulated market verticals.
CMPs manage the software image lifecycle, which covers the definition, design and deployment of commercial off-the-shelf, semi-custom and custom software platforms. When using multiple cloud endpoints, it can be a challenge to reconcile which software is in use in which cloud. Because a CMP provides high visibility into software license usage and cost containment, it becomes a powerful infrastructure management tool.
CMPs also manage quotas, which can limit the number of cloud resources an organization’s division, application team or workgroup can consume. With quotas in place, the private cloud provider ensures the availability of adequate resources for all consumers. The use of quotas also allows for reporting on cloud resource consumption, and provides a level of transparency into the processing requirements for infrastructure components. This gives consumers the ability to “shape” their usage in real-time. These consumption metrics feed into a chargeback / show back tool, which can become the basis of automated infrastructure management.
Challenges with Cloud Management Platforms
The primary challenge is the complexity of the implementation of the chosen tool, and re-design of the organization around consumption-based computing. A successful CMP implementation requires thorough planning using proven project management practices including:
- Identifying all stakeholders
- Defining user roles and establishing levels of access within the CMP
- Determining available cloud endpoints for each application or workload
- Considering the establishment of quotas, metering and chargeback / showback
One must then determine how the cloud automation and process framework integrates with the following IT support models for:
- Incident Management
- Problem Management
- Change Management
- Patch Management
- Release Management
- Configuration Management
- Image repository and lifecycle management
- Customer portal
- Usage reporting
- Configurable services (events, triggers, notifications)
- Service delivery management
As cloud endpoints are identified, greater due diligence must be performed to ensure the applications and workloads are running in the appropriate cloud end-point. Most CMP solutions allow for dynamic scaling of resources for an application. There must be consideration to the boundaries in how much resources can be added dynamically before alerts are sent that an application may not be performing correctly. Security implications must be considered if the software components require specific network ports, application service access, and generic access to physical network resources. If the solution offers dynamic building of images, the inter-compatibility of the software components that make up that image must be validated.
The cost for a comprehensive CMP can be quite expensive, with a large enterprise implementation often requiring multimillion dollar spends. While there are many products significantly less expensive, they do not offer the features and functionality desired by most organizations. The spend for cloud management is very much like an enterprise application spend, since it will touch many areas of the organization in order to support utility-based computing.
It is imperative to determine, execute and validate that all of your applications and workloads are running in their correct cloud endpoints. If the data is processed within a regulated market participant (e.g. HIPPA, Sarbanes Oxley), additional considerations may need to be factored in. For example, HIPPA regulations now require that Patient Health Information (PHI) that resides in an Amazon cloud must run in a virtual private cloud and must be encrypted both at rest and in flight. As part of a compliance program, the organization must have the ability to show evidence regarding adherence to compliance and governance requirements.
Things to look for in a Cloud Management Platform
As you can see, there are many components in a comprehensive CMP. It is more telling and important to know how a CMP will perform common functionality versus whether or not it can do it. Below are a few things to consider when choosing a CMP:
1. Which platforms / cloud provider(s) can the CMP tool support?
While you may be a VMware shop today, look for a CMP that will position you to use other cloud and platform endpoints (e.g. Amazon, Rackspace, Azure, etc.)
2. Does the CMP integrate application and system deployment and automation tools like Chef or Puppet?
The CMP should allow you to leverage and reuse current Chef or Puppet resources.
3. Does the CMP API support integration with applications to increase and reduce cloud resources based on application demand.
The CMP should allow you to auto scale an application based on conditions you define to seamlessly increase capacity while maintaining performance.
4. What level of consumption metering detail is available out of the box?
Whether the CMP has a chargeback / showback mechanism or not, the tool should provide the ability for easy retrieval of consumption of cloud and platform resources. If the Chargeback feature is not available the data will then be transferred into a Chargeback tool that the client has implemented in order to properly cost out cloud usage.
5. How does the tool allow for self-service provisioning?
One of the key drivers of cloud computing is the agility to provision resources quickly. The era of waiting 30 – 45 days to get a VM provisioned are thankfully part of the past. The CMP should support the self-service provisioning of resources with the necessary safeguards in place (e.g. quotas, resource governance and role based access)
6. Does the CMP allow for images to be ported from one cloud environment to another?
This is an effective feature for organizations that may want to perform application development in a hybrid or multi-cloud environment. Once development is complete, port the application to a private cloud for QA and production.