Whenever a company changes the name of their major product you often have to wonder what level of change they are trying to signal. In the case of VMware which changed ESX to vSphere yesterday, the signal is one of intent. They could have called it vWorldDomination but that might have been a bit too caustic. So instead they chose a global metaphor. Despite the subtlety, make no mistake, this version is a direct affront to how we have traditionally run our data centers with traditional operating systems and element-centric system management tools.
They made their case initially at VMWorld EMEA when they declared that a new “operating system” is needed in the virtualized data center and that the old model no longer applies. They called it Cloud OS but didn’t deliver on this vision. vSphere is the first step towards this new model in that it significantly shifts the focus from simply virtualizing workloads to managing and automating pools of VMs and shows how management at the virtual infrastructure layer can address data center efficiency in ways other layers can’t. It also moves the VM world closer to being able to manage business services that span VMs (although other tools like HP Operations Orchestrator and BMC BladeLogic still do this better) and track and diagnose their performance with AppSpeed, previously BeeHive, (although not as well as Hyperic).
Probably the biggest improvement in vSphere is its performance and scalability. There are a number of improvements to the maximum supported RAM and CPU configurations, as well as far better I/O throughput, and overall efficiency. In short, VMware can now claim that they can virtualize any application, no matter how performance hungry it is. Plus, VMware partners like Intel and Cisco have announced complimentary enhancements to address a variety of performance and ease of configuration issues.
Some of the key capabilities that enterprise customers should strongly consider include:
- vCenter Host Profiles -- Capture all the configuration details of a deployed virtual machine or collection of VMs and create a golden master that will enforce these configuration parameters when a service is reinstantiated, moved or duplicated.
- Thin provisioning with vStore -- Quick, what’s your storage to VM ratio today? What if you could cut this down by 90%? How many of your VMs share the same OS and middleware stack (or same applications). And your storing how many copies of that exact same software? 'Nuff said. While many enterprises take advantage of array-based thin provisioning to get some of these same benefits, vStore provides these vendors with an API to optimize thin provisioning for a vSphere environment.
- vShield Zones -- Distributed Resource Scheduling is a key technology for creating pools of resources and allowing VMs to automatically optimize their use of these resources. But managing a big homogeneous pool lets you set policies per app but not across groups of apps. Zones lets you create sub-pools for implementing these types of policies.
- vCloud -- while this is still mostly a future, VMware now has named a collection of hosting companies who plan to deploy VMware-based cloud computing in their data centers which will allow you to more easily take advantage of Infrastructure as a Service cloud computing. The big gain they bring: Your vSphere and theirs are compatible and your existing vCenter-centric tools allow aggregation of your virtual pool and their clouds.
- Fault tolerance -- for applications that require zero downtime, VMware now offers a fault tolerant capability that can eliminate the need for specialized hardware or additional clustering technologies. The fault tolerance feature replicates your VM’s state to a shadow VM on separate hardware in the same data center. If the primary machine should fail, the shadow VM picks up with no disruption to users.
But before you can really take advantage of these enhancements, enterprise infrastructure and operations professionals need to adopt more general best practices with their VMware environments such as booting VMs from network storage, activating DRS and getting comfortable with basic policies, taking greater advantage of VM templates and virtual appliances, leveraging VMware HA for recovery, and backing up with tools that leverage VMware’s backup APIs. You can’t move to a more efficiency future without preparing your environment for it.