As infrastructure and operations professionals seek to broker cloud services for the enterprise, they are coming to terms with the need to “cloudify” existing vSphere infrastructure in order to make these environments more developer-friendly and support migration of workloads to AWS or other clouds as their needs evolve.
In other words, the goal is to make vSphere look and act like a public cloud. The biggest challenges that enterprises face when trying to unify vSphere and cloud environments isn’t technical, but rather cultural. For example, getting agreement on what “cloud” even means? What cloud characteristics are required, and what gaps must be filled to cloud-enable vSphere?
There are six key requirements that must be addressed to deliver all the benefits of cloud on vSphere:
* Rapid Self-Service. According to the2014 State of the Cloud Report, 71% of cloud-focused companies provide self-service portals that enable developers to provision new infrastructure in the cloud in less than an hour. This expectation sets the bar when creating cloud-like vSphere environments.
To satisfy these expectations, organizations need to fill several critical gaps between vSphere capabilities and public cloud alternatives. Developers or even business users need to quickly and easily provision workloads through a self-service portal. Self-service also provides the opportunity for IT to deliver a catalog of standardized VMs for commonly used workloads. By offering self-service access, vSphere admins can shift from low-value activities of manually provisioning VMs to high-value activities such as offering agile infrastructure and standardized catalogs of services. A good self-service portal will also give IT teams better visibility, governance, and management.
* Migration and portability across cloud and virtualized environments. IT organizations are already working to implement hybrid IT environments that leverage vSphere and private clouds, as well as AWS and other public clouds. In doing so, enterprises are making strategic choices to avoid lock-in to any particular vendor, resource pool, or virtualization technology.
One-way migration (typically from virtualized data centers to AWS or public clouds) can provide significant value for organizations looking to eliminate data centers and move to pure public cloud services. Moving these workloads can require some refactoring of applications to operate effectively in public cloud.
However, delivering ongoing application portability goes well beyond brute force “lift and shift” migrations. There are several compelling reasons for investing in application portability including lifecycle-based deployment, cloudbursting, disaster recovery (DR), and supporting multiple hypervisors.
To maintain application portability, it is important to plan ahead and put in place a multi-cloud foundation that makes it easy to change the underlying infrastructure.
* Support for DevOps and Configuration Management. vSphere leverages black box golden images that make it difficult to track, audit, and control the software, versions, and patches included in each golden image. As a result, many enterprises create manual spreadsheets simply to track such details as the date, time, and person associated with any changes made to VMs.
Although existing vSphere golden images or VMs can still be leveraged, the process of cloudifying vSphere offers new opportunities to phase in more modern configuration management techniques. Using automated configuration management tools can help in the evolution from golden-image production to an infrastructure-as-code model. This is a key enabler for the shift to the growing DevOps development movement that streamlines development and reduces errors through automation and collaboration.
* Single pane of glass for management, visibility and governance. When virtualization first took over server infrastructure, the savings were astronomical. However, the ease of setting up virtual machines quickly led to VM sprawl, which is the habit of provisioning VMs for users with limited ability to track whether VMs were fully utilized, under-utilized, or dormant. In addition, application owners often request more compute resources than are used because it is difficult to predict in advance the appropriate level required. As a result, companies experience sub-optimal use of resources, with excess capacity often running at 30 percent.
As organizations leverage vSphere environments, AWS, and other clouds to meet infrastructure needs, they must develop an approach to managing and governing across all of the available resource pools.
Companies will continue to leverage familiar vSphere and vCenter tools to help them deploy and manage their virtualized environments. However, a cloud portfolio management solution that works in parallel with existing vSphere management tools can be used to deliver these multi-cloud capabilities. Enterprises can use cloud portfolio management to gain complete visibility and centralized governance across all of their resource pools, while continuing to use existing vSphere tools for management of their underlying virtualized environments.
* Cost and capacity management. As organizations combine both public cloud and on-premises infrastructure, they need to develop new processes and approaches to managing costs and capacity.
Prior to public cloud, organizations purchased on-premises infrastructure as discrete capital expenditures that were managed through detailed business cases followed by lengthy approval processes. While this process provides oversight on costs, it requires companies to accurately forecast capacity requirements and then either over-provision to meet peak capacity or limit application usage in order to smooth out variability in demand.
In addition, if demand was lower than expected, there was no opportunity to reduce these sunk costs, and if demand was higher, new purchases triggered another round of approval processes. Even organizations that chose to outsource data centers were typically locked into expensive, long-term contracts that often did not provide the flexibility to adapt to changing business environments.
With enterprises now using a combination of public cloud and on-premises infrastructure, they will need to master both traditional discrete capacity management approaches and continuous cost management approaches in order to minimize overall spend.
To introduce some aspects of continuous cost management for vSphere, companies will want to attach costs to vSphere resources and offer cost visibility (showback); cost tracking and analysis; and cost allocations (chargeback) for these resources. Companies should provide standard costs on standardized VM sizes running in their vSphere environments. Cost estimates should be visible to self-service users before they provision workloads, and actual costs should be visible and available for reporting and analysis at any time through the self-service portal.
* Greenfield vs. legacy applications. Cloud-enabling vSphere can help speed greenfield application development and improve efficiency in managing or migrating existing legacy applications that are already deployed in vSphere.
For greenfield requirements, a self-service portal is the critical component to helping developers leverage virtualized infrastructure for development and testing. In addition, new development offers an opportunity to experiment with or advance best practices for cloud portability, configuration management, and DevOps.
For existing legacy applications, organizations can start by leveraging “single pane of glass” management tools to gain visibility and governance across all types of infrastructure, whether cloud or virtualized. They can next identify segments of the application portfolio that will deliver significant benefits through migration to cloud, which may include mobile, social, web-scale, batch, and big data analytics applications. These existing applications can be prioritized for moving to cloud.
The cultural and technical barriers will eventually come down between virtualized and cloud environments. One of the first steps is getting on the same page on what cloud means. Armed with those agreements, everyone can get behind initiatives that move their companies forward.
Thakrar oversees RightScale’s integration with various public clouds, such as Rackspace, Windows Azure and private clouds platforms such as OpenStack, CloudStack and vCloud.