Private clouds are and will continue to be more cost effective than public clouds for organizations with revenues or budgets greater than $1 Billion. The main reasons for this are: 1) The applications running in larger organizations demand more resilience, governance and control; 2) While small businesses will find public clouds attractive, large organizations will be able to create a private internal cloud that has many of the same business characteristics as a public cloud, but with much higher levels of control, security and availability and 3) Achieving the same levels of resilience with public cloud infrastructure would be prohibitively expensive for larger organizations.
Definitions of Cloud
There are as many definitions of cloud as there are technology bloggers. However, the major theme is that a public cloud provides IT resources that are provisioned remotely from the consumer and operated by a third party. The three major characteristics of cloud are:
- IT infrastructure is abstracted from the buyers, who purchase logical compute units
- Business units see all infrastructure costs as operational (OPEX) with little or no CAPEX
- Infrastructure capacity is elastic, and can increased or decreased without financial penalty
Wikibon sees the Private Hybrid Cloud as a predominant model for businesses greater than $1B. Within this model, internal and external IT resources are logically combined to support the applications of a single organization. The users and consumers of IT resources and services see the same business characteristics as the public cloud. The IT department then funds internal cloud resources - which are built on a highly virtualized IT infrastructure - together with specific “Software as a Service” (SaaS) contracts that allow access of applications (e.g., email, SAP) from specialist organizations that take responsibility and can deliver the necessary levels of availability and security according to service level agreements controlled by the consuming organization.
In a Private Cloud, the infrastructure policies are governed by a single organization where workloads and data can be moved to and from internal and external data centers. While the external data centers can be multi-tenant in nature, Private Clouds provide an organization a single point of control for security, manageability, privacy, audit, compliance and governance.
Economies of Scale
In general, large organizations run more efficient IT departments, run more sophisticated applications, and reach more employees. Table 1 shows the cost differences in cost profile by business size. It shows that businesses over $1B in revenue spend less that $1,100/seat/year, or 38% lower than the cost/seat for a small business.
Table 1 shows more business and IT metrics as a function of business size.
Importance of Application Class
However, the cost/seat also varies greatly by the type (class) of the application. The classes of application, as defined in previous work by Wikibon, are as follows:
- Class 1 – Development, File & Print, Small Data Marts, Small-scale applications & Data Base servers
- Class 2 – Medium-scale applications & DB servers, Data Marts, Small CRM, Small Data Warehouses, Messaging (Exchange, etc)
- Class 3 – Mission Critical Applications, Enterprise CRM, ERP, Large scale OLTP, Large scale DB Servers, Large Scale messaging
- Class 4 – Large-scale applications requiring highest levels of availability, security & recoverability
Figure 2 shows how both application class and business size contribute to IT costs.
The private cloud will allow significantly improved infrastructure costs as the capability of the hypervisor software and ecosystem become enhanced as part of the private cloud journey. Class 1 and 2 applications are being migrated rapidly to the private cloud, and class 3 and eventually class 4 will migrate over time. Figure 3 shows the details of that journey.
Private vs. Public Cloud Cost Comparisons
Figure 4 give a cost breakdown for Class 3 applications for an organization with greater than $1B in revenue or budget.
The key points the data illustrates are:
- 60% of the private cloud costs are labor, and 40% are associated with equipment.
- The labor savings are limited when transferring to the private cloud.
- For class 3 applications with high integrity, security and audit requirements there would significantly increased management, audit and security costs to manage today's public cloud. Future offerings may offer higher levels of security and audit, and would shift some of the labor cost to subscription costs
- Network and network management would be more complex and costs would be significantly higher in a public cloud setting
Availability of typical public cloud offerings are 99.99% and 99.9%. This is significantly lower than the 99.99% or higher that is achieved in good large company data centers.
Figure 5 summarizes the key findings of this research:
- Public Clouds are cost effective for new Class 1/2 applications in SMB organizations.
- Private Clouds are more cost effective for the fewer very important applications for SMB organizations; a SaaS approach is likely to be the most cost effective solution
- For organizations over $1 Billion, private clouds are more cost effective in general, and provide significantly higher availability and security functionality
- There will be opportunities for class 1 development projects to use public clouds as a way of improving time to test the viability of new applications, but projects should assume that they will be eventually migrated in-house and the migration strategy and costs should be built in to the project
Action Item: The strategy for organizations with over $1B in revenue or budget should be to focus on implementing a private hybrid cloud for all applications. Judicial use of SaaS offerings will minimize the necessity for public cloud offerings. For some class 1 or 2 applications there may be a case for tactical use of public cloud offerings.
For small businesses there will be cases where public cloud will be a cost effective solution, especially for new capabilities. For existing applications virtualization in house will be an effective strategy to reduce costs. Migration costs for transferring existing infrastructure to public clouds will have to be much lower for before major transfer to the public cloud will take place.