Deep dive into VMware's virtual infrastructure
VI 3 swims through our server consolidation test, demonstrating some amazing capabilities and a few quirksFollow @infoworld
The selling points of x86 server virtualization are by now common knowledge. By moving systems off dedicated, underutilized servers, and using virtual machines to consolidate them on fewer boxes, you can reduce power, cooling, and space requirements, and you can save a bundle in hardware costs. After the bean counting, VMs can help ease provisioning, load balancing, and disaster recovery.
Less understood is the path to achieving these gains. Once you’ve caught the consolidation bug, what’s really involved in making the move, both in terms of technical requirements and physical labor? And what kind of control do you have over the new environment? To find out, we brought the heavyweight champ of virtualization platforms, VMware Infrastructure 3 (VI3), in for a deep look, subjecting the software and a supporting team of VMware engineers to one of our real-world, Fergenschmeir test scenarios.
[There’s much more to VI3 than we could fit into the space of this particular article. See also our take on important and even invaluable VI3 capabilities such as VM snapshotting and cloning, VMware’s consolidation planning tool, and how the virtualization wave is driving innovation above and below the hypervisor.]
In the end, VI3 and the VMware team passed our test with flying colors, successfully migrating a number of Windows and Linux systems and impressing us with a wealth of useful tools and automated management capabilities. We also discovered some curious limitations in VI3, however, that made our path to a virtual infrastructure a little less straightforward than it otherwise might have been.
Taking the Plunge
Our test began on a bright October morning. The first order of business was to pick a free blade in our Dell PowerEdge 1955 blade server chassis, install Windows Server 2003, join that server to the domain, and install VMware VirtualCenter Server. This installation was straightforward, with all requirement packages present on the install CD. Although there weren’t any Infrastructure 3 servers to manage yet, the groundwork was laid. Next, the first VI3 server was built on a second blade in the Dell cabinet.
Like its predecessor, VI3 is built on a Linux base, leveraging the stability and light footprint of a highly customized Red Hat operating system to provide foundation elements, but relying on a VMware kernel and VMware I/O drivers and schedulers, to squeeze the most out of the hardware. The Linux folks will immediately notice that the installer is unabashedly built on Red Hat’s Anaconda, and installation is generally as easy as booting the CD and clicking Next a few times, ensuring that the required I/O devices are discovered and configured. In the case of our Dell server, I/O was limited to one gigabit front-end NIC and one gigabit back-end NIC for iSCSI SAN interaction. Within a few minutes, the first VI3 server was booting, and the gathered geeks toasted the achievement with a brief swig of Red Bull.