BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
It's not uncommon for storage admins to have difficulty with certain aspects of managing storage for virtual environments. Tasks such as dealing with a growing infrastructure, managing storage I/O capacity and effectively using new management features can be overwhelming. In this podcast, virtualization expert David Davis gives pointers on how to get around those obstacles, and discusses some best practices for managing storage for virtual environments with SearchVirtualStorage Associate Site Editor Sarah Wilson.
What would you say is the chief challenge IT managers encounter in managing storage for virtual environments?
David Davis: To me, the chief challenge in storage for virtualization at most companies where you've got an ever-increasing workload from the virtual infrastructure is just simply dealing with that. As you replace physical servers in the data center and the new server is much more powerful, you have so much more CPU capacity and so much more memory capacity. And then you upgrade the hypervisor and it becomes much more scalable and is able to run many more virtual machines more efficiently, and the density of the VMs that you want to put on your storage continues to increase much faster than your storage. The bandwidth and I/O capacity is increasing.
In many cases, at least in my experience, we would buy a large SAN and we would put it on a five-year lease because it was something we thought would be in the data center a long time, [and] it was difficult to make that storage migration. So, you planned on it being there for a long time, but you would get a new server that had 10 times the capacity as the old server, whereas the SAN is just staying the same. So, you have to closely monitor the I/O utilization and each LUN on the storage array to ensure that that virtual infrastructure doesn't saturate out the storage I/O capacity of the SAN. So, I'd say that dealing with the cost, the planning and the complexity related to staying up with the virtual infrastructure's pace -- assuming the virtual infrastructure is expanding, as it is with most companies -- is really the chief concern for IT managers.
Additionally, I think IT managers really struggle to stay up-to-date on the latest virtualized storage advancements, the integrations we've been talking about and cloud storage options. When I go to speak at VMware user groups, that's where most of the questions come up -- around all these new features and capabilities related to cloud and further integration between storage and virtualization.
Are there any common mistakes admins make when managing storage for virtual environments?
Davis: To me, the most common mistake for storage admins is simply not embracing the idea of virtualization. In some cases, you have SAN managers or admins that have been doing their job for many years. They might be close-minded to this idea of virtualization, and in some cases they might be opponents to changing the way that things are done when it comes to managing a SAN or selecting a storage solution. So, in some cases they can slow the progress of the IT department, and even the company, to achieving greater efficiency through virtualization. I don't want to fault those storage admins and managers for trying to be pragmatic and conservative in protecting the storage infrastructure and ensuring that it's as high-performance and efficient as possible. I certainly respect their opinions and their experience, but I also encourage them to be open-minded and try to embrace this new idea of virtualization and storage, virtualization especially.
I hope storage admins will look for all the positives in what virtualization offers, which is all new ways to do backups, new ways to do disaster recovery and new ways to see the entire data center from one dashboard, one tool. No longer does it have to be multiple silos of storage, server and network. These can be broken down and can just be an efficient, fluid data center seen through one management interface. So, in the end, many of the storage issues related to virtualization are really, in my opinion, about education and communication between what have typically been the multiple silos of management in the traditional data center. So, the virtualization admins and storage admins need to work better together. They need to do whiteboarding sessions, attend the same classes, read the same books or blog posts, and really communicate about how to best make the data center more efficient by leveraging this new technology in virtualization, storage virtualization and network virtualization that's going to benefit everyone in the data center, especially the company as a whole.
What are some best practices to keep in mind when managing storage for a virtual server environment?
Davis: I would recommend to storage and virtualization admins alike to really work together to develop an application-based approach for a form of management. What I mean by that is look at the applications you support and focus on providing the best performance possible for those applications. That's really why the IT department is there, to support the company's critical applications. At many times IT people, being technologists, get focused on the technical terms and the technical features and software applications. We focus on the LUNs [logical unit numbers] or the IOPS or the [VMs], and instead we need to find ways to measure whether the end users are seeing faster screen refreshers, faster data updates, anything so that they're able to do their jobs faster and more efficiently.
On a side note, why aren't IT departments called 'business technology enablement' departments? I mean, that's what they really do. IT departments aren't just about information technology, they're there to enable the business to be more efficient, to sell more products and to make the employees able to do their jobs faster.
Something else to always look for is shared resource contention. This is something that VMware Storage I/O controller, or SIOC, works to solve -- shared resource contention, specifically storage contention, in a VMware virtual infrastructure. So, don't get me wrong, it's not a magic bullet, but simply understanding the problem that SIOC works to solve I think would be enlightening to both storage and virtualization admins because it works to solve the most common form of resource contention in a VMware virtual infrastructure.
One more thing is to make sure you map your storage and virtualization domains and create a baseline. It's always amazing to me how many large enterprises just don't have any documentation or diagrams as to how the virtual infrastructure and the storage work together. When there's a problem in the data center -- let's say an application is running slow -- they don't even know what storage that VM disk file is on, the type of storage that it is, what the expected capacity is out of it, and what the expected capacity in terms of I/Os is and how it's changed. And when that happens chaos ensues because there's no real documentation or plan. And this type of thing can even happen when you have to add a new SAN to the data center. If there's no existing documentation or diagrams or baseline, the process of adding a new SAN becomes complete chaos. So, make sure you monitor your capacity in the data center, not just in gigabytes but in I/Os because that's something that both storage and virtualization admins need to do their job. And that's really why this greater integration is happening between the storage and the virtualization -- to give those IT managers what they need, which is virtualization-aware storage and virtualization that's storage aware.