Cloud ROI is not Guaranteed

March 2013 EMA Analyst's Corner:
Six Key Considerations for Selecting a Private Cloud Vendor: How to Use the EMA Private Cloud Radar Report
Introduction: Cloud ROI is not Guaranteed
EMA research – Demystifying Cloud – has shown that a large share of private cloud projects does not turn out the way they were initially planned. Over a third (36%) of cloud projects are running late, over budget (31%), didn't meet user expectations (31%) or missed ROI goals (29%). It is important to understand that the reason for these challenges typically lies in a mismatch of the shiny new private cloud platform and the existing datacenter that has often organically grown over decades.

Core Challenge: Customer Requirements are Heterogeneous
The EMA Radar for Private Cloud Platforms (Full Report / Report Summary) constitutes an essential tool for any organization trying to identify the best cloud platform for their specific IT and business goals. Readers must keep in mind that expectations and technical requirements are often fundamentally different from customer to customer. Some organizations look for a quicker way of provisioning VMs for their development team, others aim at automating their current ITIL processes, while a third group searches for a way to cost effectively deliver virtual desktops to their staff. These are just three examples for the countless use cases of private cloud.
Six Key Cloud Considerations
There are six considerations that should be part of every cloud planning process. When reviewing the EMA Radar for Private Cloud Platforms, it is essential to carefully consider each of these criteria:
1. Greenfield or Brownfield?
EMA research has observed a 50:50 split between greenfield cloud deployments – integration with current enterprise IT management solutions is not a consideration – and brownfield clouds – where this integration is mandatory. The latter school of thought aims at utilizing cloud as a self-service layer for existing enterprise systems that enables organizations to provision and manage application environments in a more efficient manner, while the former aims at rapidly deploying IaaS environments. Both approaches come with a set of benefits and drawbacks. Greenfield clouds can be rapidly deployed, without organizations having to worry about lengthy integration projects. On the downside, the greenfield approach to cloud often requires a separate and therewith redundant set of management tools. This can lead to technology silos. Brownfield clouds require a larger upfront investment in order to ensure integration with existing systems. At the same time, the potential ROI is greater, as previous datacenter investments can be leveraged to a much higher degree than is the case for greenfield deployments.

2. Commodity Infrastructure or Blade Servers?
The traditional mega clouds – such as Google and Amazon – utilize commodity hardware that is intelligently managed by the cloud software to achieve enterprise-grade capabilities, such as high availability, bare metal provisioning and advanced capacity management. However, many organizations do not have the capabilities to create and manage a cloud that is based on commodity hardware and guarantees corporate SLAs. The use of blade servers or even converged infrastructure – pre-integrated servers, network and storage – takes significant complexity out of the cloud environment, dramatically simplifying management and, as a result, has become more and more popular within the previous years. On the downside, converged infrastructure solutions lead to vendor lock-in and can generally increase hardware cost.

3. One Hypervisor or Multiple Virtualization Platforms?
Recent EMA research has shown a strong correlation between cloud adoption and organizations deploying multiple hypervisor platforms. This is due to private cloud platforms treating hypervisors as a commodity, by offering features such as workload portability, high availability, backup, snap shots and policy based workload management. Taking advantage of multiple hypervisor platforms enables customers to benefit from cost, density, and performance differences between hypervisors. For example, organizations may choose to run critical production workloads on VMware's ESXi platform, while hosting less critical environments on the free KVM hypervisor. While a multi-hypervisor strategy can lead to significant cost savings, organizations still must have the skills to maintain and upgrade each platform.

4. Open Source or Commercial Platforms?
There are generally two ways of taking advantage of open source cloud platforms.

a. Free Open Source: The organization directly downloads CloudStack, OpenStack or Eucalyptus and molds the selected platform to the corporate requirements profile. In order to ensure a successful implementation of an open source cloud, the customer must have significant cloud experience and the ability to customize, maintain and support the code base on an ongoing basis. Every organization needs to determine whether this commitment makes sense compared to the adoption of a commercial cloud platform.
b. Commercial Open Source: The customer selects a commercial distribution of the preferred OpenSource cloud technology. Citrix Cloud Platform is a commercial distribution of CloudStack, while IBM, HP, Dell, Cisco, Nimbula, Morphlabs and PistonCloud offer well supported OpenStack solutions. In addition to the free open source version of Eucalyptus, the company also offers a commercial license with enterprise-grade support. Selecting a commercial distribution of an open source cloud platform enables customers to still benefit from the code released by the open source community, while also receiving enterprise-grade support from a major vendor.

The risk of committing to any kind of open source platform – commercial or free – is that this platform may fork and the adopted version may become less popular or even abandoned.
5. Virtualization Vendor or Big 4?
VMware and Microsoft are the vendors owning the largest share of the hypervisor marketplace, while the Big 4 – IBM, HP, CA Technologies and BMC Software – dominate the marketplace for automation and management. Both groups of companies offer their own cloud solutions. Opting for a Big 4 cloud solution often makes sense for organizations that already are Big 4 customers or have adopted multiple virtualization platforms. In this case, adding a cloud layer to the existing IT management tools can unlock efficiencies, while keeping implementation cost low. Opting for one of the virtualization platform vendors, such as Microsoft or VMware, may make sense for organizations with significant existing investments in either company's virtualization technology. On the downside, vendor lock-in always is a concern when creating a monolithic cloud platform.
6. SaaS focus or IaaS/PaaS?
While SaaS may be the ideal delivery vehicle for entire enterprise services, the required upfront investment may be too large for many customers. In order to deliver desktop, server and public cloud applications through a SaaS cloud, the organization must "wrap" these applications with deployment and management instructions. The reward for this labor intensive process constitutes of centrally managed application environments and business services that can be easily and consistently deployed in an ITIL compliant fashion. In the end, every organization has to determine whether it has the capabilities to create a SaaS cloud or if IaaS or PaaS may be the better solution.
Readers of the EMA Radar Report for Private Cloud Platforms should combine these six considerations with a detailed analysis of the Radar profiles of the vendors they are most interested in. While the bubble chart presents an overview of the marketplace in general, this chart does not constitute a replacement for a thorough analysis of how the reader's individual project requirements match the vendor strengths and limitations outlined in the vendor profiles.