Debunking private cloud myths

Many businesses considering transitioning their business to the cloud do not like the idea of being on a shared model. Enter Private Clouds!

With private clouds customers have their dedicated compute, storage and network environment that is secure, scalable, redundant, cost optimized and operationally efficient. Unfortunately, many IT professionals think that moving from a virtualized Data Center to a private cloud environment is a simple task involving a few management software and automation. It is necessary to debunk this myth and throw light on general misconceptions about moving to a private cloud environment.

Three common myths about private clouds that this feature will try to bust are:
Automating virtual machines for cloud is a simple task
Provisioning storage for private clouds is easy
Configuring network is simple

By definition private cloud resides within the data center of the organization or at a third party facility. It allows complete control of IT resources, automates workflows by eliminating manual intervention for shifting workloads, provisioning compute, storage and other IT resources, etc. One can move application workloads, reallocate memory, storage and other IT resources, and consolidate data and management in a single location.

Automating Virtual Machines for cloud is a Simple Task

Most businesses today are on virtualized platforms where virtual machines (VMs) can be provisioned for various services that the organization runs. Within a cloud environment (read Private Cloud), application owners should be able to provision VMs on-demand - a completely automated mode so as to reduce operational workload of administrators and thereby eliminating possible human errors while provisioning VMs.

Creating, Provisioning and Managing VMs

More and more organizations are taking the private cloud route after toying around with advanced virtualization for a while, so that they can provision VMs without having a team of people to do it.

There is a stark difference in creating, provisioning and managing VMs between virtualized environment and cloud. Let us see what that is.

In a virtual infrastructure, there is a clear process for creating VMs so as to avoid VM sprawl - VM templates are created with standardized hardware and software settings that will include base OS, service packs and other patches. To ensure that there are no performance issues and other management issues, IT departments have not installed full-blown applications and services into VM templates.

In a cloud environment the idea is to make the provisioning and management of VMs a user-driven process. The goal is to allow end users to create new applications and services on-demand. Users will have to be given a complete application and service catalogue and not just base OS. VM templates will have to be loaded with applications and services - basically work closely with applications owners and application managers to create various VM template configurations.

The challenge here is to have the VMs tested and verified well, put in place required controls to ensure that VM sprawl does not happen. The risk of empowering users will have to be balanced with proper controls to ensure corporate standards of performance and optimized use is maintained.

A tiered approach by creating classes of VMs where IT managers have pre-defined various hardware and software configurations including various applications and services can help in controlling performance and optimal usage. One can also set chargeback or showback policies to various departments within the organizations so that users are aware of their resource utilization.

Application owners should be able to provision VMs from the private cloud on-demand just as they can with an external provider. This can help cloud administrators reduce operational issues of provisioning VMs, managing the virtualization layer and automating the provisioning and managing of VMs. The challenge is to ensure proper cataloguing to ensure reduced customization; implement proper controls to reduce VM sprawl and optimize usage by various users.

Create separate environments for Development, UAT (User Acceptance Testing) and Production. Start with development and UAT and then move production to the cloud environment. It is also important to put an expiry on provisioning of VMs so as to control VM sprawl.

Provisioning storage for private cloud is easy

Provisioning and management of storage in a cloud environment is not as easy as the infrastructure itself.

Making storage working in tandem with private cloud infrastructure has some challenges associated with it. First of it stems from the difference in the evolution of both virtualization and storage technologies.

Typically in a virtualized infrastructure environment working with a storage area network (SAN) will have direct access to set up and manage storage zones or pools. Whereas in a cloud environment, the virtual server acts as a single image with processing (CPU), RAM, storage (working as direct attached storage) and applications and services embedded as a single unit. Individual server along with a SAN device works very differently from this. To make things even more complicated, there are equal chances that the storage within this environment is also virtualized.

Thin Provisioning enables over allocation of storage for increased utilization, improved uptime of applications and simplified storage capacity management. In reality what happens is that the IT teams lose sense of actual usage and within the cloud environment it can create chaos.

It is important to clearly identify the various ways in which the infrastructure and storage interact. Linking a two virtualized resources into an infrastructure agnostic platform is not going to be as simple as it sounds.

There is a paradigm shift in the process of technology management that is an even bigger challenge. In traditional virtual environments access to storage was stringently controlled through technology as well as process of access approvals. In a cloud environment users can provision storage on-demand easily which leads to a new challenge - how to manage storage utilization.

While the actual management and policy for cloud-enabling storage is challenging, it also is a boon in disguise. Cloud managers can make provisioning and use of storage easy alongside reinforcing the importance of optimized utilization through chargeback or showback policies.

It is important that cloud managers remember to avoid dual overcommit. Use of capacity planning tools can ensure a synchronized balance between allocated capacities to utilized capacity.

Configuring Network is Simple

Provisioning and managing of networks within a private cloud environment is a challenge - the same network of a traditional virtualized environment will just not do for a cloud-ready infrastructure.

How will bandwidth allocation work in a cloud-ready network? This is a key question. In a managed virtual environment where the network administrator did the management of bandwidth, the cloud infrastructure will create new challenges. As users get more freedom to create and manage VMs, there is a high risk of network congestion due to users setting up VMs and creating massive loads, which end up choking the bandwidth.

In a private cloud environment, most processing happens at the data center; unlike at both client and server sides in a client-server environment. This means continuous communication between data center and the users, which can quickly turn into a bandwidth allocation nightmare.

It is important here to identify and segregate resource hungry users, have fair-usage-policies in place, create backup bandwidth to accommodate increase in loads, etc. In a virtualized environment it was only consolidation of physical servers into lessor numbers, which did not hinder network traffic, but in a private cloud environment you are offering more services and empowering the users to do more with the infrastructure. This will have huge impact on the network traffic, and understanding, identifying, taking corrective actions are some starting steps. Upgrading the network bandwidth and allocation policies is almost a necessity.

It is also important to understand and design the network to take advantage of the essence of private clouds. For instance, it is important to keep network consistent when moving between workloads. Future technologies such as SDN (Software Designed Network) can help achieve this.

Now coming closer to the compute environment within the private cloud setup. The virtual switches also need some consideration. Traditional virtualization environment was provisioned with virtual switches that connected to the VMs. Most virtualized environments are done with static model for assigning of ports to VMs. Here is the next challenge. In a cloud environment the number of ports can quickly run out in numbers. There will have to be a dynamic way of assigning ports and destroying them as and when they are needed or otherwise.

In Conclusion

Although cloud providers tout ease of private cloud implementation, what IT leaders have to understand is the readiness of the existing setup. No doubt the benefits of a cloud environment are unbeatable and unquestionable, it is important to have a partner with proven experience to help and migrate to private cloud.

True essence of private cloud can only be achieved if all elements of the infrastructure are part of the whole strategy - this includes network, applications and services, people, etc.