Virtual changes

Roger Baskerville, vice president EMEA of Vizioncore Inc. moves beyond physical and financial storage limitsVirtualisation has changed a lot about the way IT systems are architected and provisioned, but the effective allocation of available storage remains unsolved.
    
There is plenty of guesswork involved, even for the most experienced administrator. But guessing wrongly can have severe implications: assign insufficient disk to a virtual host and performance will suffer and the server can fail; assign too much and resources will be unused or unavailable to the virtual machines that need them. In an environment with potentially hundreds of virtual machines, that unusable disk adds up to one very large, very expensive problem.

Adding a buffer
So how does capacity get trapped inside a virtual environment? As noted, under-provisioning storage for VMs – just like in the physical world – results in serious performance problems. So most administrators tend to follow recommendations to over-allocate the storage available to VMs. Additionally, virtual machine templates add a buffer of additional storage resources to prevent performance issues. All these buffers add up.
    
At the low end this could be 30 per cent of unused storage. At the high end this can be as much as 70 per cent of unused storage. Multiply this figure by hundreds of virtual machines in even a medium-sized environment, and you have a formula to really exacerbate excessive storage use and cost.
    
A single server housing 20 virtual machines and corresponding applications can potentially overwhelm the storage resources and units assigned to it. Unlike other virtualised resources, such as processors, unused storage allocated to a virtual machine is not available for others to share.

Tackling the challenge
There has tended to be three distinct approaches to meeting virtual storage management challenges. The first has been to reduce the virtual machine size and resulting storage needs through compression, defragmentation and de-duplicating. The second is to give users the ability to dynamically reallocate storage and meet anticipated demand.
    
The benefit of these approaches is that reducing of the virtual machine size helps preserve available memory, whilst reallocation helps shift it to where it is needed. But there has been a third option, and it is pretty straightforward. Buy more capacity; either in the form of physical assets or virtual appliances such as IBM’s Storage Volume Control (SVC).
    
Compression and reallocation has not solved the problem, as it only serves to reallocate available stories. No provisions are made for the trapped storage that is available for use. As companies have virtualised, a new approach based on the unique characteristics of this technology has emerged that enables the dynamic allocation of disk resources to- and-from the resource pool.

Resetting the machine

The only questions for any administrator who wished to tackle this job are how much time and how much virtualisation know-how do you have? The answers to those questions are needed to make the best decision on what kind of approach to take to resolve the issue. Most administrators believe that a solution that can easily and automatically reset virtual machines and their storage buffers to their optimal sizes and return any space that is freed to the SAN or data store fore reallocation, represents the strongest value proposition.
    
With such a solution, what options remain for resizing virtual machines? On VMware’s ESX 3.x it is almost impossible to shrink an existing virtual machine as well as the guest OS NTFS and partition. But it is possible to enlarge this manually, which can take up to a couple of hours to complete per virtual machine. This process will have to be done sequentially. It is easy to see how resizing even 50 virtual machines could become a daunting task.
    
Added to this is that it is manual, meaning the process is a risk to continuity because of the potential for human error. In addition, software tools often need to be purchased to help business complete the resizing process, which can be far from easy to understand.
    
Shrinking a VM down means the administrator requires deep knowledge about the VMDK files inner structure and the use of HEX editing. So if it needs to be done manually, it is going to be necessary to create a VM backup copy before performing any manual resizing.
    
Another benefit of using an automated resizing solution for reclamation is that it really supplements existing storage management methods and can represent the final component of a well-rounded storage management approach. Here is how a storage reclamation solution would complement the most common approaches to virtual storage management:

Saving storage costs
Integration with hypervisors and management applications – the reclamation solution can be used to automate VMware ESX Server-based VM partition adjustments. Unused VM space can be reduced by 50-80 per cent, resulting in substantial saving of virtualisation storage costs because more VMs can be hosted on each ESX Server. An optimisation tool that fully integrates with VMware vCenter Server can be used to automate storage monitoring, resizing and reclamation process.
    
Support of thin provisioning – a storage reclamation solution can be highly effective when used to complement current thin provisioning tools. Optimisation solutions find wasted space, unnecessary applications, utilities and temporary files within each VM; optimises the VM size; recovers unused storage and releases the unused storage back to the thin provisioning application for reallocation.
    
Integration with block storage virtualisation appliances (eg. SVC), such a virtual appliance, provides more storage space for virtual environments. Reclamation solutions provide separate functionality and can work alongside solutions like SVC, so users can make more efficient use of the additional space that is provided.
    
Solutions that recapture unused storage truly represent a unique and new approach to virtual environment management. Organisations have traditionally addressed their storage needs by either reducing VM sizes and adding capacity (via virtual appliances or physical assets), or they use provisioning to play musical chairs to move resources where they’re needed most.
    
These approaches dramatically increase support requirements and add unnecessary capacity acquisition expense to the virtual environment. However, optimising VMs and reusing saved space allows users to get more out of the resources they already have. This is hardly a new idea, but it is only now that it can become reality thanks to the ongoing innovations that virtualisation technologies enable.

For more information
Vizioncore’s products help customers safeguard and optimise their IT systems while allowing them to extract the maximum return on their investment in virtualisation. For more information please visit www.vizioncore.com or www.vizioncorum.com, Vizioncore’s official blogging site.