Managing storage for virtual environments: A complete guide
A comprehensive collection of articles, videos and more, hand-picked by our editors
There are a slew of tools available designed to ease enterprise storage management in a virtual server environment, whether they're products or services provided by third-party vendors or features of the virtualization hypervisor. More recently, storage virtualization and cloud technologies have been recognized for their impact on storage management. But do all of these approaches really make enterprise storage management easier? In this podcast, expert David Davis explains to associate site editor Sarah Wilson whether virtualizing storage and cloud services help enterprise storage management, and provides examples of some of the best tools available to storage administrators for simplified management.
What are the benefits of virtualizing storage within a virtual server environment?
David Davis: Just as server virtualization was able to reduce the number of virtual servers in the data center, these new virtualized storage solutions may be able to eliminate storage arrays completely. If they can't completely eliminate a storage array -- because that's a huge task in a large enterprise data center -- they'll be able to make those storage arrays so much more efficient when it comes to providing storage for the virtual environment. What this means is that you'll be able to save money through the efficiencies that are gained. Even if you can't completely eliminate an enterprise storage array, you gain benefits like per-virtual machine [VM] replication through your storage array because it can now understand the virtual infrastructure; you'll gain per-virtual machine statistics through the storage perspective, which is very important, and then features like super-fast cloning, snapshotting and much more.
And then there's another type of virtualized storage feature that I haven't yet talked about, and that's an exciting new group of products that are server-side caching for virtualization. They come from companies like Infinio Systems and PernixData, and what these do is they take SSDs [solid-state drives] in many cases -- you don't always have to have SSD in your physical vSphere host -- but in many cases they utilize SSD to provide massive performance increases for your VMs, and then of course whatever applications are running on those VMs. In effect, they give you much greater VM density on each of your vSphere hosts, and also on your storage arrays. So the idea, which I really like, is taking what you already have and just increasing the performance and scalability of it -- and that's what they call server-side caching solutions.
What about cloud services? How can those fit into a virtual server storage management strategy?
Davis: Of course, every company under the sun is talking about the cloud, and I don't want to completely throw that out just as market-speak because there are some very valid solutions. Some of them are like second-tier storage for primary data, like the TwinStrata cloud array. What it does is have a virtual appliance that you install, and you start storing your virtual machines there -- and whatever data you want can be stored on this virtual appliance. Then it begins replicating that data to the cloud and it acts as a high-speed cache for the most common data -- that data is stored on the virtual appliance. And so it's kind of like an infinite cloud array. So you have unlimited capacity as it's stored in the cloud. So I think that's a really innovative solution, but it's not a solution that I see most companies adopting or most companies interested in.
The more likely solutions that I think you would see companies be interested in would be backup as a service and disaster recovery as a service (DRaaS) solutions. And that's really the best place for enterprises to get started with cloud computing, at least in my opinion. If you could send your backup data off-site to the cloud instead of using some arcane tape drive that you have to wait for to write to and put in boxes and have someone come pick up to take offsite, and then worry about their reliability if the tapes should fail or get lost -- if you could just replace that with backup to the cloud, it's just a much more simple and beautiful solution. And the same for disaster recovery: If you could have a DRaaS solution in the cloud so that at any time your VMs are already replicated there, you could just flip a switch and start using those VMs and that data instead of your primary data center when a disaster occurs. That's just a much simpler solution than maintaining your own off-site secondary data center, replication solutions and so forth.
I think if a company can afford those solutions, and they can test them and they're a good solution for that company, I think it's something that should definitely be considered. Just to name a few examples, backup solutions that have cloud integration come from Veeam backup and replication, PHV Virtual and there are many others. And then for DRaaS, for example, you can use Zerto and VMware [vCenter] Site Recovery Manager, with a bunch of different cloud providers. There are a lot of great solutions out there, so I just encourage you to do your homework and try to think past all the cloud-related marketing campaigns.
Let's talk about what products or services are available to make enterprise storage management easier. What do you think are some of the most helpful tools for virtual server environments?
Davis: The tools that you use for managing storage and virtual storage, in my opinion, should be the same. It should just be a storage tool whether that storage is virtualized or that storage is on a traditional enterprise storage-area network. Now, you also have virtualization management tools, so you kind of have two different categories of tools there: virtualization tools and storage tools. But both of these categories of tools are getting smarter, and they are able to understand the storage if it's a virtualization tool, or the virtualization if it's a storage tool. So they're becoming more aware of one another.
One example is in these new virtualization-aware storage arrays where you can see per-VM I/O statistics. To me, that's a huge advancement because if you can make storage decisions based on I/Os per VM, you'll know where to place that VM, you'll know where your hot virtual machines are -- if you want to move those virtual machines into an SSD-based LUN or something like that, a hybrid LUN with greater performance or faster caching -- that's something you can do. Then in your virtualization tools, if you can quickly sort all your VMs by IOPS and see how many virtual machines are on a LUN, how many I/Os that LUN is serving, then I think you're doing well and you're using some smart tools there. Traditionally, these types of statistics that are related to I/Os per VM have been difficult to get and the enterprise has typically turned to third-party tools. For example, vCenter Operations Manager has some storage views to help you keep up with storage issues in virtual environments. You can also look for tools that are traditionally storage tools that support SMI-S [Storage Management Initiative Specification], which is the storage monitoring protocol, and those can talk directly to the storage and dig even deeper. They can get really deep information related to performance, utilization and capacity out of the physical storage arrays, but you also want to make sure those tools can understand your virtual infrastructure.
I think more and more of these types of management tools are becoming one -- they're bridging the gap. You have virtualization tools that use SMI-S, you have storage tools that talk to the virtualization layer, and it's just becoming a matter of selecting the right tool for your company.
What about some of the most useful storage management features supplied by prominent hypervisor vendors like VMware or Microsoft. How can those help with enterprise storage management?
Davis: Both VMware and Hyper-V support the movement of VMs, virtual disk files, from one virtual machine to another with no downtime. In VMware vSphere that's called Storage vMotion, and in Microsoft Hyper-V that's called Storage Live Migration. VMware has the unique features of VAAI, vStorage APIs for Array Integration, and they also have VASA, which is the vStorage APIs for Storage Awareness. And Microsoft has ODX, which they call Offloaded Data Transfer. All of these are great ways for the hypervisor and the storage to work better together.
I think they're great steps in the right direction, but most of these features are relatively 1.0 features that immediately provide some benefits. But there's lots of room for improvement, and I think many more advancements will come.
Storage DRS, VMware's Distributed Resource Scheduler, is also very innovative. It will automatically move running VMs from one data store to another. But I'm not really sure how many enterprises have deployed it in production because it's going to take some time for them to get a comfort level around their virtual disk moving from one array to another array during the day with end users using it -- applications are actually hitting that storage while this virtual machine disk file is transitioning across the network. It's going to take some time for them to feel good about that and just click the checkbox to enable it on full automatic mode, and I’m sure it's the storage admin who's going to feel the most discomfort about doing that. Still, it's a very innovative feature and I think it's a peek at the types of innovations that virtual storage can provide us in the future. There are also advanced graphical enterprise storage management tools out there, but so many VMware admins still monitor storage, believe it or not, with tried and true command line tools like ESXtop.