Solution provider takeaway: Server repurposing creates an on-demand virtual infrastructure and provides key benefits to a virtualized environment. Solution providers can use server repurposing to improve their clients' VM migration efforts.
A virtualized environment is inherently an on-demand one, right? If your customer needs another server, you just create another
Unfortunately, you most likely didn't explain this to your customers before they bought into virtualization. You presented virtualization as a way to maximize near-idle compute resources to reduce physical server counts, thereby reducing power, cooling and management costs. But you probably failed to mention that adding physical hosts to the system in a seamless way would shrink the savings they gained by virtualizing in the first place -- that's because in order to easily accept new hosts into a virtualized environment, you need to have excess compute capacity.
Going forward, to avoid such a disconnect between expectation and reality, you should design virtual infrastructures to have plenty of headroom for spikes in compute demand and for bringing new virtual machines into the system. Unfortunately, this will result in extra power and cooling costs in the data center.
The ideal virtualized solution would require fewer hosts than your virtualization customers now have; it could run at near-maximum capacity but have the ability to power on additional virtualization hosts as needed. Having an idle bank of hosts connected to the network and storage but in a powered-off state means that when a peak load hits, you can simply point at the powered-off bare-metal machines, power them on, connect them to the right storage and network resources, and then boot the correct virtual OS image. Once booted, you could shift over the virtual machines that you need to the new virtual host, thereby rebalancing your customer's compute resources.
Server repurposing products from companies such as Scalent, Egenera and Cassatt can enable such a scenario. With these solutions in place, you can power on physical hosts that were previously powered down, point them to specific machine images and then boot the appropriate OS, virtual or otherwise.
These products address another big problem in most virtualization environments: the inability to easily remove a server from the virtualized environment. Say, for example, your customer has three physical hosts: two running at about 50% of compute capacity and the third at 25%, and then another server outside of the virtualized environment running a critical standalone application. If the nonvirtualized server starts to fail and the application is virtualization-friendly, you could use a physical-to-virtual (P2V) tool to bring the application into the virtual infrastructure and rebalance resources. But if the application is very high I/O and needs to be on a standalone server, without a server repurposing application, your only choice is to reformat and reload the OS and application.
Server repurposing applications give you another choice when a high I/O application demands its own server -- you can shift the virtual resources from the third machine to either of the first two servers, unvirtualizing the third machine, and then moving the physical image from the failing machine to the newly freed-up third server, now outside of the virtual environment. A server repurposing application also gives you the ability to restore the original configuration once the failing server is replaced.
With a server repurposing system in place, it would take about as long to make these shifts -- adding new physical servers to the virtualized environment and migrating machines in and out of the environment -- as it takes to boot the new machine or OS image. As a result, server repurposing tools make excellent second-tier availability solutions for applications that can handle five minutes or so of downtime, offering an availability capability similar to that provided by VMware HA.
The flexibility and speed to redeploy lets these products save more costs for your customer, allowing them to further optimize computer resources and increase power efficiencies. Because the next virtual host is only a point, click and OS boot away, there is less risk in stacking up virtual machines on fewer physical hosts. The result is an environment where your customer has fewer physical hosts running at much higher levels of compute capacity and with idle, ready-to-go hosts powered off. When a compute spike occurs, a simple point-and-click powers on the previously powered-off host, loads the OS, and then virtual machines can be rebalanced. When or if the compute spike subsides, you simply reverse the process. And this process can be automated by templates that provision server, network and storage resources in the correct order.
In addition to server repurposing, these solutions enable virtualized network connectivity. Newly powered and booted servers can be virtually attached to any network and storage resources in the data center without recabling. Servers that are connected to the lab network and lab storage for testing, for example, can be pressed into action if the need arises. That server simply needs to be pointed at the virtual image, network connections and storage connections established and the virtual server farm is extended in the time it takes to reboot the lab server. There's no need to carry the server to the correct rack and manually connect it to the right network, eliminating the chance for error and saving the time needed to physically move the server. To make all this work, of course, your customer needs networked storage and servers that boot from it, which can bolster your pitch for centralized storage.
By presenting server repurposing to your customers, you can greatly help them increase the effectiveness and flexibility of their virtualization project through the creation of an on-demand virtual infrastructure.
About the author
George Crump is president and founder of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. With 25 years of experience designing storage solutions for data centers across the United States, he has seen the birth of such technologies as RAID, NAS and SAN. Prior to founding Storage Switzerland, George was chief technology officer at one of the nation's largest storage integrators, where he was in charge of technology testing, integration and product selection.
This was first published in September 2008