As virtualization technologies have emerged as a means to drive down cost and increase data center efficiency, the role of technology as a competitive tool has fallen by the wayside. While virtualization offers real benefits in terms of cost savings, improved server performance and minimized storage requirements, it also provides network administrators with essential flexibility and agility in managing data center environments.
In terms of a company's competitive advantage, data center efficiency is a key consideration, and efficiency is sometimes expressed as a "time-to rating." A time-to rating measures how quickly a data center can implement, respond and recover from changes. From an infrastructure perspective, this means virtualization. Figure 1 shows the
For any layers of the virtualization stacks, infrastructure virtualization can have a significant impact on the time-to capabilities of companies. With a price tag that's typically less than $1,000 per server, infrastructure virtualization enables data centers to virtualize messaging and storage networks and to control physical servers connected to these networks. In terms of a time-to capabilities, infrastructure virtualization delivers rapid deployment, rapid adoption of change and flexible disaster recovery (DR). Vendors such as Scalent Systems Inc. and Unisys Corp. have addressed data centers' time-to needs with solutions that provide immediate payback not only on the time-to scale but also with cost reduction and power reduction.
Scalent's and Unisys' infrastructure virtualization offerings enable data centers to react to changing business needs in real time, by making the transition among different software, network, and storage access configurations or the transition from dead-bare metal to live connected servers in the time required to boot a server and without physical intervention. Infrastructure virtualization software allows companies to implement cost-effective business continuity and real-time automation while reducing server counts, simplifying manageability and increasing reliability.
The software typically installs on an x86 server, logically sits off to the side of a data center and plugs into the top-level switch. The software discovers the relevant hardware automatically. A backup image of each server needs to be created on a storage area network (SAN) or network-attached storage (NAS) device. Then there's an applet to install on each server backup image. With that, the physical servers network boot the appropriate image and the infrastructure is virtualized. Typical installation takes about one to two days including the creation of the backup images.
As mentioned, infrastructure virtualization enables you to image a server in the amount of time required to boot that server. Also known as server repurposing, this allows an administrator to redeploy a server with a new OS or new application in a matter of minutes by simply clicking a button. This type of redeployed OS can be any OS and at any time; however, it is not limited to nor does it require a server virtualization OS like VMware. The IT infrastructure can have a few powered-down bare-metal servers on standby as a result of infrastructure virtualization. When a shift in the business architecture arises, redeployment or recovery can occur quickly, efficiently and seamlessly. When expanded beyond a single physical machine to ten or even hundreds of servers, server expansion takes no longer than deploying one additional server. It remains a point-and-click process, and a sys admin need wait is only for boot time.
As another example, if a request is made to deploy a new Microsoft SQL infrastructure, 10 cold servers can point to 10 template images. In a matter of minutes, those 10 servers can be deployed and ready. If the number were 50 SQL servers, it would require a few more clicks but no additional time. If storage virtualization were added to this example, storage can be provisioned from the servers. A new virtual domain emerges with the capability of some of those systems. In a matter of minutes the entire environment can go from cold servers to a fully deployed system with provisioned storage.
Demonstrating even greater flexibility, if a request is made to move an application to a new hardware platform, merely pointing the new bare-metal server to the current server image and initiating a power-on sequence is all that is required. Conversely, keeping the current server hardware but deploying a new storage image involves only copying the server image to the new storage and assigning the server to it. In both cases, the infrastructure virtualization technology handles mappings and assignments for the new messaging network and storage network seamlessly.
Time to recover is particularly critical in VMware environments where running in a high-availability mode is not enough protection. Failure of a single VMware node means a massive redistribution of virtual machines (VMs) throughout the VMware architecture, or overloading the VMs from a failed ESX server to an ESX server that is going into maintenance or to another machine. Infrastructure virtualization can boot a new cold server, load the ESX image on that server, and handle all the network and storage mappings while maintenance is performed on the original ESX server.
Should access to the primary site be lost, time to recover is critical. In this case, it might be acceptable to move non-virtualized machines into a virtual environment to save on idle server resources at the disaster recovery site. With an infrastructure virtualization tool, it is possible to move the physical environment to a smaller virtualized environment at the DR site including all the correct SAN and local area network (LAN) connections. This enables "N to 1" failover, reducing not only the time to recover to the DR site but also the cost of idle servers at the recovery site. In fact, if the DR site is another office in the organization, these servers can be used for testing and development until the eventual disaster hits, eliminating idle-server costs. Infrastructure virtualization allows for changing the mode of a server in no more time than it takes to boot a server.
Infrastructure virtualization offers data center management capabilities in terms of, provisioning, automation, failover and virtualization:
- Real-time server failover
- Reduced hardware for test infrastructure
- Eased deployment and management of hypervisors (real-time physical-to-virtual and virtual-to-physical, removed storage and network constraints)
- Real-time server and associated network and storage connectivity provisioning
- Data center power management
- Cross-data-center disaster recovery-oriented failover
- Cyclical application server consolidation
- Automated redeployment of servers from physical platforms or from physical to virtual environment for consolidation
In sum, infrastructure virtualization provides essential agility and disaster recovery capabilities and substantial competitive advantage for those companies ready to meet the challenge. As you make the case for virtualization, don't allow the clear potential for cost savings and efficiency eclipse the equal benefits of data center agility, dexterity and company competitive advantage that virtualization clearly offers.
About the author: George Crump, founder of Storage Switzerland, is an independent storage consultant with over 20 years of experience.
This was first published in March 2008