Data Center Knowledge | News and analysis for the data center industry

Cloud Sizing and Deployment: a Look at Best Practices

Cloud environments will vary greatly based on corporate demand and existing IT resources. The goal of any organization is to use IT at its most effective level. For an IT manager, this means planning is going to be crucial. Cloud sizing doesn’t have one correct answer, since every environment is truly unique in its demands.

However, there are tactics in making a cloud initiative successful, so let’s look at a few considerations and best practices.

Identify the Needs

This step is absolutely crucial in deciding what physical and virtual components will go into the cloud. When determining needs, administrators must look at a variety of business drivers.

What will the user be accessing? How will the user connect to the cloud environment? How many users is the cloud platform going to support? Working with maximum capacity requirements and knowing where more resources are needed are all a part of this step.

When the need is established, cloud sizing can actually occur. If the need is miscalculated, core cloud components may be either under-sized or even over-provisioned.

Growth Considerations

As discussed earlier, cloud environments must be ready for growth. As the demands on any IT environment increase, administrators must be ready to respond with available resources in their cloud environment.

This doesn’t mean having idle resources just sitting on VMs or physical hosts. It means planning each server for capacity and having extra for usage spikes.

As user count grows, existing hardware should safely handle the workloads until new cloud devices are provisioned. At that point, the user load can be load-balanced between cloud servers as needed.

Current and Future Capacity

Proactive environment planning goes beyond just having a few extra resources on a physical box. “Future-proofing” a cloud environment means planning for expansion based on business needs.

To accomplish this, IT managers must be in sync with key business partners. If an IT executive doesn’t understand the organization’s business drivers, there is no way that they’ll be able to size and balance the cloud environment properly. This type of proactive infrastructure capacity approach will help with business growth and create savings by having an environment ready for expansion.

Physical Resource Allocation

When the need is established, administrators are able to provision physical gear into their cloud environment. Even when working with third-party providers, IT managers must know how much resource allocation has to happen.

In a robust infrastructure, administrators know the capacity of each server and its ability to handle a given workload. These administrators are able to proactively load-balance users to servers with fewer loads to continue operating optimally.

As more resources are required, the environment should be able to handle users above its designed capacity as a form of failover. If an emergency happens, other servers must be able to handle the load of a failed host. This type of planning revolves around good visibility into the cloud environment.

Deployment Methodologies

Cloud deployment methodologies will differ as well. As mentioned earlier, it will all depend on which environment (or combination) is being used. When deploying a successful cloud environment, it’s important to truly understand the following components:

Workload type. This means knowing what is being deployed to the end-user. Are we pushing a small application with limited requirements, or is a full virtual desktop being rolled out through the WAN. By knowing and understanding the workload, administrators are able to plan their cloud environment accordingly.
The WAN link will always be a concern. In an earlier section we covered the type of cloud environments available and the type of bandwidth they will likely need. It’s very important to remember that proper bandwidth planning can mean the difference between a robust cloud environment and major latency issues due to WAN congestion.
Current and future resource needs. Planning for the future will create an environment ready for growth and expansion. IT managers need to establish a baseline for their current needs and then understand what their environment will be doing one, two, and sometimes three years from now. The more information gathering revolves around resources the better the environment can scale in the future.
User training and acceptance. Although cloud technologies may be understood by many IT professionals, this concept is still very confusing to the average user. Administrators must take the time and logically explain the environment to the end user. This means training, webinars, and easy-to-follow documentation. Even the best cloud environments can fail, if the user doesn’t know how to properly utilize the infrastructure.

Cloud Use Cases

One other important cloud sizing point is knowing how the environment will be utilized. Since there are many ways to use a cloud, we’ll cover a few popular methods for cloud utilization:

Cloud HA and workload automation. Many organizations use cloud deployments as a means of augmenting their existing datacenter environment. For example, an organization specializing in travel might use cloud services to quickly spin up new VMs capable of handling additional user load during peak seasons. This strategy can prove most effective, especially since there is no need for additional infrastructure purchases.
Cloud DR. Disaster recovery is a popular cloud initiative. Depending on the type of DR plan, organizations are able to create active/active or active/passive environments. In fact, entire data center workloads can be replicated and made available should a disaster even occur. Remember, with disaster recovery, there are entire additional dynamics with planning and resource management.
Cloud expansion. Organizations that expand fast through acquisitions leverage cloud technologies to quickly integrate new business entities. By simply providing a web-facing cloud portal, the new business unit is able to access the centralized workloads offered by the parent company. This fast and efficient type of cloud provisioning will help organizations get online faster and without the need for additional infrastructure costs.

Remember, there is no one way to properly deploy a cloud infrastructure. This is mainly because the needs of a business organization will always vary.

During the early phases of a cloud planning project, it’s important to include all necessary business units and cover the topics discussed above. Whether the cloud environment is being used strictly for DR or for a remote branch, the needs of that infrastructure will always be relative to the demands of the organization.

This means that the more data you gather prior to deployment, the better your cloud environment will operate both now and in the future.