Interpreting Private Cloud Deployments

Article by Karan Kirpalani

Technology has long matured to help build and take full benefits of private cloud but the adoption has been rather slow

According to reports done by leading analyst firms on enterprise technologies, global private cloud adoption figures claim to be over 30% with over 20% claiming to be moving rather fast on the build-out. A deeper look at these reports will reveal the reality of the true adoption of private clouds. The very definition of Private Cloud is skewed in most cases – and in most of the mentioned 30% cases the key features of private cloud such as metered usage, multi-tenancy, etc. are clearly missing.

When it comes to private cloud implementations in the enterprise space, the interpretation itself is the most clouded.

Myth: Virtualization is Private Cloud

The most mistaken technology set is right here. Virtualization often gets confused with private cloud. The fact is that virtualization is only a component of cloud – Private or Public.

The intent of virtualization is to merely increase utilization of many servers for a particular workload, using logical partitions created over the same physical device(s).

A Virtualization solution is implemented to provision infrastructure faster and consolidates servers – done by using a hypervisor and set of management tools. The management software provisions Virtual Machines (VMs) where OSs can be installed and workloads can be run – and depending on the workload that has to be run, configurations of compute, storage andnetwork access needed, are provisioned accordingly.

Orchestration software is used for provisioning, and the virtualized infrastructure is connected to the local network that manages the VMs.

Truth about Private Clouds

Private Cloud, on the other hand, is a whole different ballgame when it comes to increasing utilization of IT resources. It leverages multi-tenancy to differentiate workloads into ‘tenants’ that are used at different times allowing one to consolidate utilization of all IT resources across the data center – meaning very high utilization of existing resources.

Private Cloud is a totally different scenario although virtualization forms the basis of all cloud buildouts. The key five characteristics of a private cloud are:

  1. On-demand Self-service
  2. Broad Network access
  3. Resource Pooling
  4. Rapid Elasticity
  5. Measured or Metered Usage
Private Cloud Virtualization
On-demand Self-service Self service portal with service catalog and automated provisioning Manual Provisioning of VMs for end user requirements
Broad Network access Service accessible via internet or local network Available only for users authenticated on the enterprise network
Resource Pooling Resource sharing (compute, storage, network) from underlying hardware through virtualization Yes
Rapid Elasticity Dynamically increase or decrease IT resources provisioned Manual provisioning of VMsincluding installation of OS and apps – restricting flexibility and faster scalability
Measured or Metered Usage Monitoring and reporting service usage for chargeback on actuals Charge back on traditional budgeting methods – not on actual usage

Cloud should be viewed as an API through which you can control capacity and provisioning of resources, such that you can automatically move workloads.

A Case in Point

Now, visualize a scenario of a centralized IT department of a large organization that supports several branches and departments withspecific IT requirements at various times. In order to provide better support for their branch offices and various departments without hiring more staff, fast turnaround time when provisioning service for applications, they virtualize the IT environment and deploy management tools in a way that applications and OSs are preinstalled in disk images.

Depending on access levels, users can now perform tasks such as launch and install VMs from supported images. A branch or a department wanting to try out a new application or a tool can now provision a new server (read VM), install the application and use the same for a period of time specified during provisioning. Such deployment of varied workloads, all happening in parallel, add to the overall efficiency and productivity levels of an enterprise.

True Sensing Private Clouds

This rapid turnaround, without having to put in capital budget for new servers and,ability to turn services on and off as and when required(read servers in this case) is really the true sense of a private cloud – the IT department is using virtualization, but they have added a layer of self service for branches/departments (On-demand Self-service) that can access services over a local network or a virtual private network (VPN) over the Internet or a web-based interface (Broad Network Access). The users, depending on access, can provision IT resources quickly and turn the same off equally fast (Rapid Elasticity) with little help from the IT department, and without having to go through the process of raising a capital expenditure for the resource. IT department can meter actual usage for each of the service or provisioned resource by the branch or department and charge back on an actual usage basis (Metered Usage/ Measured Service).

In Conclusion

The important factor for organizations to consider is their goals of building a private cloud.What an organization wants from a private cloud depends on what business they are in; one size fits all strategy may not work here. Providers of hosting and managed services as well as cloud service providers can help assess and identify the size and shape that fits these organizations.