NAS clustering: What it is and how it's being used

Find out about the different meanings of the term NAS clustering, how it's being used for scalability, its limitations and how it differs from file virtualization.

The precise definition of NAS clustering -- which generally refers to the grouping of NAS resources -- can differ depending on who you talk to. And, there are other related terms, such as scale-out NAS, that are sometimes used interchangeably with "NAS clustering." In this podcast interview, Greg Schulz, founder and managing partner of the StorageIO Group, discusses the different types of NAS clustering. He also defines its limitations, explains when a third-party data migration tool might be needed for use with NAS clustering and describes how it differs from file virtualization.

You can read a transcript of the interview below or download the MP3.

Play now:

Download for later:

Listen to the Podcast
• Internet Explorer: Right Click > Save Target As
• Firefox: Right Click > Save Link As What is NAS clustering?

Schulz: NAS clustering means a lot of different things to a lot of different people. In one instance, it's considered clustered if you simply have a pair of NAS heads, or nodes, in an active-passive configuration. If one fails, the workload shifts over. That's more analogous to the classic server cluster definition. But if you go by that definition, almost every block storage device on the marketplace could call itself clustered because they're all at least dual-control for failover.

[There's another kind of] clustered NAS -- some [people] call it scale-out NAS, some call it bulk storage-type systems -- where you have multiple nodes that you can move file systems across, that you can load-balance, that you can increase your performance and capacity for a given file system.

Where it gets even more confusing is clustered NAS vs. a clustered file system, where your file system can be running on multiple nodes at one time. Of those groups, which is getting attention in terms of new product releases?

Schulz: It's all over the board. Certainly, if [you're talking about] two or more controllers for failover, those are very commonly deployed, [so there's] a lot of action around that. If [you're thinking] more in terms of scaling out to boost performance, boost capacity, in addition to increasing availability, that's where you're seeing a lot of activity in the area of supporting bigger file systems, larger home directories, online archives, active archives, repositories [and] hosting Web 2.0 applications. What are the limitations of NAS clustering?

Schulz: The biggest limitation can be the different approaches. There are some that are very proprietary, and there are some that are very open. Some are so open that you could take the software that typically comes prepackaged on a server and plug it into different vendors' storage systems. So you have a lot of flexibility, maximizing your return on investment. Some are those deployments but prepackaged, preconfigured in a solution where you don't have as much flexibility. You start looking at more solutions that are focused on bulk, scale-out, high capacity, ultra low cost for capacity. You may be giving up some of the enterprise features such as snapshots, advanced replication and other functionality. What about migrating data using NAS clustering?

Schulz: If you go to a clustered file system, where the actual file system is running across the different nodes, by design, once you put your data into that file system within that cluster system, the data is moving within the file system itself. So you simplify the moving, the re-tiering, the load balancing. Even if it's not in the same file system, [with] most of the clustered solutions, once you put the data within the cluster, there are tools for migrating between one file system and another. In terms of how you get your data from an existing NAS onto a clustered file system, [there are] a lot of tools out there. If you have it as an NFS mount point or a CIFS share today, do a copy [or] do a backup and restore. You may run into a scenario where you may want to do that while data's being used, nondisruptively. That's where you may take a look at some of the different migration tools or file virtualization tools, like those from vendors like EMC, among others. How does NAS clustering differ from file virtualization?

Schulz: Ironically, file virtualization is very similar to block virtualization. It's the same value proposition addressing the same issues -- you can use it for consolidating, you can use it for pooling, aggregation, but you can also use it for agility, flexibility, re-tiering, helping facilitate migration. They can be [complementary]: If you have a clustered NAS, with many different individual file systems on it, you may need to put a file virtualization layer across the top of that [for] ease of use. That file virtualization could be as simple as Windows DFS … or a third-party virtualization tool. But the other approach to it is if you go into a clustered NAS with a clustered file system, you eliminate the need to put that abstraction layer across all the individual file systems because you potentially have one large file system or the file systems are all contained within one system. So, on one hand, a clustered file system can eliminate the need for file virtualization but file virtualization can also complement clustered NAS and clustered file systems.

Dig Deeper on Primary and secondary storage