The Problems with VVOLS

For simplicity, most storage administrators create a finite number of LUNs for their VMware Infrastructure and assign as many virtual machine (VM) datastores to each LUN as possible. Management is simplified thanks to a limited number of LUNs but it makes storage management of the individual VM more difficult. VMware administrators need a finer grain of control than the LUN to manage better storage I/O performance. VVOLS is VMware’s attempt to provide this granularity. While VMware’s attempt is noteworthy, it is only a half step to automation. The problem is there is no middle ground between manual management and automation. Also VVOLS, quite obviously, take a myopic, VMware only view of the data center. The data center needs a more universal and complete implementation of automation, a VVOLS Everywhere.

As we discuss in our article “What Are VVOLS?”, VVOLS are another component in VMware’s software defined storage (SDS) strategy. They provide administrators with the ability to set performance policies on a per VM basis. While VVOLS are a step in the right direction, they are just a first step in addressing the many needs of the entire data center.

SDS solutions may take a more comprehensive approach to solving the problem, but many of these solutions are also LUN centric and have limited or no granular understanding of data. Many SDS solutions also lack VVOLS’ ability to create service levels for certain volumes and move data between those volumes based on policy. SDS needs to advance to address the broad performance and capacity management of the entire data center beyond just LUNs and VVOLS. VVOLS has its questions, and Software Defined Storage 2.0 may have the answers.

What is SDS 2.0

As Storage Switzerland discussed in a recent article “What is Software Defined Storage 2.0?”, SDS 2.0 extends the first generation of SDS solutions by adding quality of service (QoS) and automation functionality to the technology. It is also an ideal extension to the concepts that VVOLS introduced. But more importantly it extends the VVOLS concepts and makes them more appropriate for the entire data center.

The Limitations of VVOLS

VVOLS Requires VMware

VVOLS is a 100% VMware solution. That means for it to be universally valuable to the data center, the data center needs to be 100% virtualized, and in the case of VMware it has to be 100% vSphere. While there is tremendous value in achieving high levels of virtualization, few data centers will achieve a 100% virtualization level, and fewer still will be 100% virtualized on a single hypervisor.

Virtualization alternatives like Docker, OpenStack as well as direct hypervisor competitors like Hyper-V are capturing the attention of the traditional data center. The data center of the future may be 100% virtualized or containerized, but it is highly unlikely that it will be 100% VMware. The mixed reality of the modern data center is where SDS 2.0 comes in. It provides VVOLS-type functionality, and more, across a broad range of operating environments. SDS 2.0 is essentially VVOLS Everywhere.

VVOLS Still Requires Volume Provisioning

While VVOLS limits the number of volumes that an administrator has to create, it still requires the creation of volumes. Volumes are organized based on the type of performance that is expected from those volumes. Storage is not treated as a single object that is subsequently parsed out from there. The SDS 2.0 treatment of VVOLS Everywhere allows storage to be a single object. The administrator just sets a capacity limit and a performance requirement; SDS 2.0 does the rest of the work in the background.

VVOLS Requires Performance Analysis

Part of the challenge with VVOLS is that it requires that the administrator understand the performance capabilities of the storage that they are assigning as they create their performance tiers. In the simplified architecture of one all-flash array, one hybrid array, and one hard disk array this is not an impossible task. But the real data center may have multiple instances of these array types making analysis more complicated. Further compounding the analysis is server-attached flash storage, which the administrator may want to leverage as well.

Of course, the problem is that IT professionals are already stretched too thin in terms of responsibility. The reality is that because of time constraints, many administrators will be forced to make the best guess in terms of performance classification.

SDS 2.0 self-analyzes all the available storage resources, both local and shared, using the best available resource at the most appropriate time. With SDS 2.0, assigning storage to a VM, bare metal server or container is simply a matter of defining a capacity limit and a performance expectation. The SDS 2.0 solution will do the rest of the work automatically.

VVOLS Does Not Automatically Move Data

While vSphere’s Storage DRS’ automatic mode will re-position workloads based on performance demands, most users are not comfortable with having an entire datastore moved automatically. They want some level of control over that process. As a result, most VVOL changes in performance will be done manually by the administrator in response to user complaints.

SDS 2.0 addresses the hesitation to move workloads automatically by making the movement of data more granular. SDS 2.0 only has to move the actual part of data that needs to have its performance problems addressed, not the entire volume. A granular handling of data makes data movement far less impactful to the overall infrastructure and can resolve changes to performance demands almost instantly. This lower impact allows administrators to trust SDS 2.0 to take action automatically. There will also be situations where administrators will want direct control over data placement. SDS 2.0 allows human override of the automation, but this is the exception instead of the rule.

SDS 2.0’s performance granularity also allows it to be more efficient with premium priced storage resources like flash. Since only the sub-sections of data that need high performance are moved, the use of these resources is more efficient.

Conclusion

VVOLS is a step towards creating a storage infrastructure that is agile enough to respond to the needs of the business. It is however just a half step. The next generation of solutions, SDS 2.0, will provide a VVOLS Everywhere experience that spans multiple hypervisors and environments. It will also be more efficient, moving data at a sub-volume level. For environments that are more than just VMware, SDS 2.0 provides the answers to the questions that VVOLS creates.

Sponsored by ioFABRIC

Eight years ago George Crump, founded Storage Switzerland with one simple goal. To educate IT professionals about all aspects of data center storage. He is the primary contributor to Storage Switzerland and is and a heavily sought after public speaker. With 25 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS and SAN, Virtualization, Cloud and Enterprise Flash. Prior to founding Storage Switzerland he was CTO at one the nation's largest storage integrators where he was in charge of technology testing, integration and product selection.

Tagged with: , , , , , , ,
Posted in Article

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 21,520 other followers

Blog Stats
  • 874,262 views
%d bloggers like this: