With virtual servers taking over the data center, planned storage systems must reap the benefits of virtualization.
Server virtualization is quickly becoming the norm in data center storage. The ability to host several operating systems and applications on a single server cuts capital costs for new equipment, and the reduction in physical servers helps trim the power and cooling, floor space and number of people required to keep them running. In many areas where power is limited, power and cooling can be the primary driver for adopting server virtualization. With power limitations, the only way to add new applications or grow existing ones is to find new efficiencies. Beyond power conservation, virtualization offers some very compelling features focused on high availability and data mobility.
But with the benefits of virtualized servers come new challenges to managing storage attached to a virtual server environment. I'll suggest steps you can take to better prepare for your virtual environments. But first let's take a look at what's happening in the data center.
The new data center
There are a number of trends shaping the next-generation data center, including:
Pressures to increase utilization. Companies are always trying to find ways to reduce costs in addition to power and cooling savings. Many have found that a lot of waste in the data center could be eliminated by increasing utilization and thus forestalling new hardware purchases.
Enabling higher levels of service. The ultimate goal for IT is to deliver a higher level of service to the business. This entails rapid provisioning of applications and guaranteeing higher levels of availability. With more businesses trying to adopt a Web 2.0 approach, IT needs to develop a "2.0" infrastructure to accommodate the increased need to share information with greater ease.
As server virtualization moves from the test lab to production deployments, storage managers need to deliver at least the same level of service. The problem is that server virtualization creates a layer of abstraction between the virtual hosts and the storage. To get through that layer of abstraction, IT needs tools that can see into these virtualized environments and provide the detailed information required to make intelligent decisions about them. While server virtualization can reduce the number of physical devices in the server environment, the number of applications--and the storage required for them--remains the same.
Storage and virtual servers
Storage managers should carefully consider how they'll handle the following challenges associated with managing storage within a virtualized server environment.
Visibility or the lack of it. To effectively manage your environment, you need to be able to see it and measure it. In the past, it was easy--you had one application on one server and as long as the management software could see the host bus adapter (HBA) in that server, you could correlate the infrastructure to the application. But it becomes much harder when you have applications on a virtual host that can easily transition from one physical server to another. Keeping track of the application and establishing the proper connectivity and redundancy is vitally important to maintaining high availability.
Heterogeneous environments. The server virtualization market is currently dominated by one player, but others are gaining traction. With no standards in this space, support will be on an application-by-application basis. While efforts are underway to create standards, these are typically multiyear efforts.
Performance implications. How will you ensure optimized storage performance for virtualized applications if you can't establish the link from the virtualized environment to the physical storage? We're talking about the ability to correlate applications to the storage and being able to not only visualize but collect performance metrics from the application to the storage. It's challenging enough in nonvirtualized environments, but with multiple applications running on a single physical server, this kind of detailed analysis is critical for determining high-growth applications and resolving performance issues.
Troubleshooting. When there's a failure in the data center, determining the actual cause of the outage can be time-consuming and costly. This is also true in nonvirtualized environments. The key to accurately troubleshooting an environment lies in understanding each device and its dependencies on other devices. When an event occurs, SNMP traps are sent from not only the devices that have failed, but from all the devices in that path. This flood of alarms must be sorted out and correlated to determine the most probable cause.
Networked storage. To take advantage of some of the high-availability and data movement features in server virtualization, the virtualized servers need to be on networked storage. For many, this will mean augmenting an existing environment. For others, it will entail designing and building a networked storage environment. Decisions will need to be made about the type of networked storage--NAS or SAN--and if it's SAN, whether it should be Fibre Channel (FC) or iSCSI.
Steps to take
There are some issues storage managers need to consider when server virtualization is deployed, but none of them outweighs the benefits of virtualization. Like any new technology that becomes prevalent rapidly, there's some lag time before the industry adjusts or develops tools to leverage these new capabilities. The following are some things storage managers can do to prepare for managing storage in a virtualized server environment.
Connectivity. If your company plans to use high-availability and data mobility features:
- Check your existing SAN environment to see how much excess capacity it has. Do you need to expand your existing FC SAN or will you consider deploying an iSCSI SAN? Is your NAS environment capable of supporting the storage requirements from all of the servers that will be consolidated? Does the IP network have sufficient capacity?
- If you don't currently have a networked environment, will you deploy one? Plan for enough time to evaluate and test systems that meet current and future needs. Check with your users to determine future requirements.
Management tools. Some available storage resource management (SRM) products support server virtualization.
- If you have SRM software, talk to your vendor to see if it supports (or plans to support) virtual server environments. Ask which technologies it plans to support and when.
- If you don't have SRM software, investigate the products that support virtualized server environments. Make sure the products support your storage environment as well as your planned/existing server virtualization environment.
Manual solutions. Lastly, there's the age-old manual-correlation effort. This consists of filling out spreadsheets to track the actual path from the virtual host to the storage. This is a time- and resource-consuming exercise that will be outdated as soon as it's finished, but it only requires existing personnel and spreadsheets.
The momentum surrounding server virtualization can't be ignored because the benefits are too great. The best way to adapt to this change is to make sure you're prepared for it. And the best way to start is by understanding and embracing the value of a virtual server world, while keeping in mind all the potential problems that can (and will) develop. If you can't see it or find it, you can't fix it.