The time has come to jump onto the cloud computing bandwagon, but for many organizations that still means the need to maintain control of IT environments and physical resources. Often, legislation or regulations prevent
Even though each IT environment is unique, there are a few best practices for taking your private cloud project from planning to production, including choosing the right hypervisor, software and hardware as well as appropriate WAN and bandwidth technologies.
Hypervisor choices and software licensing. Cloud technologies revolve around virtualization. The entire idea of cloud computing is built around efficiencies and agility. The first piece of technology to look at is the hypervisor, and there are three major vendors competing for the top spot: VMware, Citrix and Microsoft. No single hypervisor option is best. Each product has its own comparable feature set; picking the right hypervisor will depend on your organization’s goals.
Once IT managers choose a hypervisor, they must decide which resources to deliver via the cloud. This decision involves several complex licensing and software considerations.
When working with Microsoft, for example, you need to determine a licensing model that works best for your user base. In some cases, an enterprise licensing agreement might be the best option. Use an experienced cloud partner that specializes in licensing to help you pick the right licensing level.
After choosing a hypervisor, there are additional considerations for building a private cloud, including:
- Application virtualization tools
- Quality of service software tools
- Monitoring software
- Operating systems
- Replication software
- Other business-related software suites
Content delivery methods. The topic of delivering workloads is a very important one that should be asked early in the private cloud planning process. How will you deliver cloud-based workloads to end users? How IT teams answer this question will often dictate several hardware and software requirements at end points and at the private cloud.
The most important part of a private cloud deployment will be the end-user experience. And end-user acceptance can be the “make or break” aspect of a private cloud’s success.
Users will connect into a private cloud to access applications and data; cloud architects must set up a strategy on how access is granted to those cloud users. For example, if end users are accessing a private cloud-based virtual desktop that’s hosted on XenDesktop, how is media latency regulated? Is content rendered at the end-user level or on cloud servers? What is the latency threshold?
Cloud admins must answer these questions when determining the best delivery method. The most important part of a private cloud deployment will be the end-user experience. And end-user acceptance can be the “make or break” aspect of a private cloud’s success.
Hardware options. Cloud computing is built around efficiencies; therefore, hardware decisions must be made with efficiency in mind. Switching, storage, monitoring and network wiring all play important roles in creating a robust private cloud.
One big hardware question IT teams have when planning a cloud environment is: Should I use blade servers or rack-mount servers? The best answer has to do with what you’re running. Since each environment is unique, selecting the proper hardware depends on requirements of the application and the business. Some cloud managers prefer a blade server environment while others swear by rack-mounted servers.
Companies looking for agility and the ability to scale quickly might want to go with blade servers. A chassis can hold several blades, which requires minimal configuration. When additional users are added to the environment, a blade server can be provisioned quickly to take on the load.
On the other hand, organizations that know the end-user environment will not grow can work with rack-mount servers. Because the servers don’t have to scale quickly, working with a set number for a longer period of time is sufficient.
There are a few other hardware decisions you must make before launching a private cloud, including storage sizing, networking configurations and hardware monitoring:
- SAN sizing: Take the time to understand your cloud environment’s needs; it will pay dividends
in performance later. A few questions to ask when building a private are: How many users will be
accessing the system now? How will user numbers change a year from now? What types of applications
and databases will the organization use in the next 12 months? Good storage
sizing practices can alleviate storage performance bottlenecks down the road.
- LAN considerations: Be certain you have the right LAN connectivity going into a private cloud
environment. This means testing for Fibre Channel and Fiber Channel over Ethernet (FCoE) support as
well as testing networking speeds. Does the environment require a 10 GbE infrastructure? An
organization can have some of the best servers available in the data center; however, if the
switches can’t handle the traffic, performance can take a serious hit.
- Monitoring: There will be times when you’ll need to use third-party hardware monitoring tools to manage and maintain a high level of service in the private cloud. Monitoring the private cloud must be ongoing to maintain the environment’s optimal performance.
WAN and bandwidth technologies. Depending on the size of your enterprise’s cloud infrastructure, you have several bandwidth options. To establish your bandwidth requirements, you need to know how many users will access the environment, what resources you will deliver via the cloud and what peak use might look like (time of day, load usage, application needs, etc.). In many instances, organizations would choose among MPLS, optical circuits or carrier Ethernet services for their WAN connection.
By following these best practices, cloud admins can design and launch an agile private cloud. The goal of any cloud initiative is to create a robust environment that scales in line with business growth and IT needs.
Bill Kleyman, MBA, MISM, is an avid technologist with experience in network infrastructure management. His engineering work includes large virtualization deployments as well as business network design and implementation. Currently, he is a virtualization solutions architect at MTM Technologies, a national IT consulting firm.
This was first published in March 2012