As the resource demands for virtualization workloads increase, converged infrastructure solves the growing compatibility, scalability and performance issues that white box servers present.
More on converged infrastructure
Navigating through converged infrastructure hype
Integrated IT stacks: Offering your own bundle
Virtualization hardware and data center design
On one hand, the advent of cheaper and more powerful servers has given IT more purchasing freedom. But building white box severs and assembling data center hardware through a piecemeal fashion can also create performance bottlenecks, which converged infrastructure can eliminate.
In virtualization’s early days, processing power and memory greatly limited performance. But vendors are slowly resolving many of these early performance issues by delivering powerful, cheaper servers. Because of the price, adding servers into a monthly operational budget requires fewer signatures than before. But purchasing cheap servers and components for white boxes may lead to issues down the road.
New build process, same inefficient white boxes
The process of building white box servers has evolved over time, but these servers can still contribute to serious performance bottlenecks in a data center.
Building white boxes fell out of practice in our industry. It seemed like a good idea in the short run, but it ended up creating long-term operational problems. Ten white box servers sported ten different hardware configurations, ten sets of drivers and ten times the chance that an update would cause a server failure.
Today’s data centers are plagued by a close cousin of the white box approach: buying servers piecemeal. Many virtual environments today are comprised of “whatever server made sense during that purchase.” Often, IT shops trust virtualization’s resource abstraction to handle different server configurations. In the process, they created another generation of inefficient data centers.
Converged infrastructure targets the performance bottlenecks
Thankfully, hardware vendors have responded and developed converged infrastructure products that address the inefficiencies of white box servers.
Converged infrastructure is just a fancy name for “hardware that’s designed with virtualization in mind.” For example, Hewlett-Packard Co.’s VirtualSystem VS1 for VMware offers a combination of hardware and software that is preconfigured and tested to ensure functionality as it scales. Converged infrastructure products, such as VirtualSystem, enable IT pros to purchase, and later scale, an entire virtual environment in a rack.
Depending on the configuration, HP’s product delivers a rack of equipment that can include enough A5800-24G switches, ProLiant DL360 G7 and DL380 G7 servers and storage space within a P4500 storage area network to support a measurable load of VM workloads.
Converged infrastructure components require software support to abstract hardware resources into resource pools. These pools function in a similar way to the resource pools found inside virtualization platforms, such as VMware vSphere. But converged infrastructure resource pools offer flexibility for scaling the hardware infrastructure as your needs grow.
The lack of seamless scaling with the piecemeal approach has been a primary source of virtualization-related data center failures, and converged infrastructure products target those problems.
Migrating off white box servers
Our industry innocently believed that hardware abstraction through hypervisors could easily handle any collection of servers, networking and storage. Abstraction does allow VMs atop any hardware to look and feel the same, but its limitations lie in VM migration and resource expansion.
VM migration has long been a hidden pain point in many virtual environments. Simply put, some hosts cannot live migrate virtual machines to other hosts because of incompatibilities between different processor types. Converged infrastructure’s hardware intends to eliminate, or at least minimize, these incompatibilities while maintaining high availability.
In addition, converged infrastructure’s combination of hardware and-software intelligently mitigates the resource expansion problem. HP’s VirtualSystem VS1 uses the HP Matrix operating environment for orchestrating the activities of servers, networking and storage. The HP FlexFabric Network then logically handles the I/O between physical and virtual machines.
HP isn’t the only player in the converged infrastructure market, which also includes Dell, Hitachi and Cisco Systems. Together, these vendors offer similar tools and varying levels of maturity, which give you the opportunity to migrate off those white boxes for good.
This was first published in March 2012