When it comes to a VMware deployment, it's all in the prep work. The more upfront work an IT staff does, the more likely it will deliver a scalable, redundant virtual architecture that will grow with the organization for years to come. Use these tips to get your project in gear.
1. Think strategically.
Virtualization works best when it's considered a core piece of an organization's IT strategy. This requires executive buy-in at the highest levels. Don't think of VMware as a point solution that will consolidate servers. Think of it as a platform for supporting desktop replacement, disaster recovery and even future integration with cloud services.
2. Focus on project planning.
Depending on the size of your IT environment, devote six or more months to planning. This phase should include a full study of your servers to determine which ones will make good virtualization candidates.
Most mainstream servers, such as file and print and even e-mail servers, are good places to start. But not every product supports virtualization. In some cases VMware may not support a specific product, or the server manufacturer may not support virtualization. Sometimes configuration requirements, such as high disk I/O and memory and large CPU requirements, make for poor virtualization candidates. Be sure to ask for a data center virtualization candidate study as part of any contract.
3. Dig deep into the many features of virtual servers.
Most IT managers fantasize about starting with a clean slate: an empty data center where they can build an infrastructure from scratch. Virtualization offers that opportunity.
If you think strategically, you can also virtualize network switches and create new configurations that may have been beyond your budget in the past. For example, many of the latest VMware versions let IT departments create firewalls between virtual machines and monitor network traffic within the virtual infrastructure for easy port identification.
Beyond servers and switches, there are new monitoring tools, as well as desktop virtualization, live server migrations from one server to another and a myriad of network configuration options not previously available. One feature lets IT dedicate network segments for virtual machines in specific security classifications. The IT team can also use 802.1Q virtual LAN tagging on physical connections and let virtual switches assign that segmented traffic to specific VMs.
4. Remember to plan for storage.
Determine your storage hardware's capabilities, especially if you're thinking of using the virtual infrastructure for disaster recovery. Make sure the storage hardware is certified for VMware and that the configuration is stable and current.
Although VMware doesn't interact directly with the underlying storage configuration, it does depend on a well-planned storage design. For example, letting VMware snapshots store data in the same logical unit number (LUN) as your VMware virtual machine files can cause issues. If unchecked, they can fill up the LUN, triggering outages.
Most modern hypervisors are storage-protocol agnostic, but make sure you have in-house expertise in the storage technology you select. The configurations for storage can differ drastically depending on the product and protocols you select. Splitting storage between operating system disks, data disks and snapshots can be helpful. You can take this approach further by separating storage between replicated and unreplicated data as well.
5. Look beyond server virtualization.
Server consolidation is usually considered the starting point for a virtualization project. If your planning, design and deployment are correct, you can use this infrastructure as a building block for additional virtualization efforts.
Most IT shops look to virtualize desktops and disaster recovery following a server virtualization project. After that, the sky's the limit: You can move on to application thin provisioning and delivering cloud services.