Today I'm able to offer another guest-authored post on Virtual Volumes, this time from my respected colleague at Dell, David Glynn, a highly esteemed Technical Solutions Engineer in their Enterprise Storage division who has spent a great deal of his time recently focused on Dell's implementation of vVols. Here David shares Dell's vision of the benefits of vVols, both to the storage administrators and the vSphere administrators: Explaining how this shift in the way we deliver storage is truly transforming the industry.
August 24, 2015
Virtual Volumes (vVols) is the biggest change to the vSphere storage stack since VMFS became a clustered file system with ESX 2.0. This new functionality really changes the way storage is used in virtualized environments — and not just with vSphere, but across the industry. VMware is once again shaking things up in the data center, and Dell Storage is already taking advantage as one of the first vendors to support vVols, starting with our PS Series arrays.
So why all the excitement around vVols? In a word, granularity. Granularity of array capabilities for storage policies, granularity of data protection, granularity of metrics. While these features are nothing new, they lacked flexibility because they were previously applied at the datastore level. Now, vVols adds flexibility and granularity by transforming virtual machines from being a series of files sharing a datastore with other virtual machines to being a series of volumes directly on the array. This results in vSphere becoming storage-aware, and storage become VM-aware.
Another word that comes to mind is licensing. Often, we see a great new feature released by VMware, but as administrators, our excitement is short lived because the feature is tied to a higher license level then we have. However, because vVols is provided with the Standard license, this is a clear signal that VMware intends for it to play a significant role in how storage is consumed by virtual environments. VMware has also moved other features down to the Standard license. The VAAI APIs, which offload tasks to storage and leverage array intelligence, and the MPIO APIs, which are used by the PS Series Multipathing Extension Module to reduce latency and increase bandwidth, have both been greeted with excitement by Dell Storage customers.
So what is meant by storage becoming VM-aware?
This can be interpreted in many ways, but what it boils down to is this: An individual virtual machine is now a series of volumes on the array, rather than just one of many virtual machines sharing a volume. This helps the vSphere administrator because many storage-intensive tasks can be completely and efficiently offloaded to the storage, which is better placed to perform tasks such as:
• Cloning a virtual machine, and therefore deploying a virtual machine from a template, which benefited greatly from VAAI Full Copy, now becomes a simple volume clone operation that is completed in seconds.
• Snapshotting an individual virtual machine, and more importantly recovering a virtual machine from a snapshot, uses the array’s efficient pointer-based snapshots, enabling snapshots up to 32 levels deep to be taken in seconds and kept forever — or long as you have storage capacity.
• Recovery, which is just as easy because the array simply brings the snapshot online as a volume.
• Array-based replication, which will in time extend to this per-virtual-machine granularity and will be supported by Dell Storage.
All of these capabilities are nothing new from an array perspective — snapshotting and cloning volumes are Storage 101. What is new is the direct interaction between vSphere and the array to the benefit of individual virtual machines.
In addition to the storage becoming VM-aware, vSphere is becoming storage-aware. Not only is the vSphere environment gaining from the improvements mentioned above, but the array capabilities advertised to Storage Policy Based Management have gone from being a fixed string of capabilities assigned to an entire datastore, to a more granular “pick the capabilities of interest to you and apply only them to the virtual machine or virtual disk”. This level of granularity provides much more flexibility in the use of storage policies (each of which contain a defined set of capabilities) across the variety of storage that can exist in a vSphere environment. This enables vSphere administrators to apply the most suitable storage profile to a virtual machine as it progress through its lifecycle. After all, a virtual machine and its application do not have the same storage performance, SLA, or security needs in development as they do once they are deployed to production.
While these are great benefits for the vSphere administrator, how does the storage administrator benefit from storage being VM-aware? It would have been simple to treat a vVols datastore as any other volume, but as a virtual machine is now a series of volumes on the array, this provides great opportunities to display additional information to the storage administrator. At Dell Storage, we never pass up an opportunity to make our customers' lives better, and information is everything. With vVols, we now enable the storage administrator to clearly see the how much space is allocated to a virtual machine, and how much it is currently consuming.
More importantly, the storage administrator can also view storage array performance with granularity down to the individual virtual machine and its individual virtual drives. This enables both vSphere administrators and storage administrators to have a clearer view and a better understanding of potential performance issues in their shared environment — resulting in quicker resolution and less finger pointing.
Another way that Dell Storage strives to make administrators' lives easier is through easy-to-use interfaces, and for vVols, we took it a step further. All of the administration tasks necessary for configuring and deploying virtual volumes — everything from access controls to the creation of vVols datastores — can be completed alongside our existing datastore management, snapshotting, and replication functionality for VMFS volumes. Each of these tasks can be performed within the vSphere Web Client using the latest release of Dell Virtual Storage Manager version 4.5. After all, the software-defined data center is about blurring the lines between traditional data center roles.
This is only the first release of vVols from VMware, and while it does not change the industry overnight, it does shake things up. Rest assured, Dell Storage will be there with VMware along this journey. Development of the next generation of vVols on Dell Storage is already underway, and we plan to add support to our SC Series arrays next year.
Take the excitement you see about vVols today, scale it up and replicate it, and you’ll be one step closer to seeing what the future will bring. We look forward to driving this data center transformation with VMware.
My thanks once again to David and the folks at Dell Storage for contributing to this blog.
If you're at VMworld 2015 in about a week, you can catch David presenting on Dell's vVols implementation at STO4805 - How Virtual Volumes Will Solve Storage Challenges in the Datacenter and on our vVol technical panel session STO5522.