By design NAS is incredibly simple networked storage. It's easy to set up: just turn it on, set up the file system, mount it on the servers or desktops that will be using it, and it's ready to go. NAS is also effortless to operate, provides easy data protection with snapshots and mirroring, and is painless to manage with some caveats. These caveats are related to managing multiple NAS systems, because there are limitations on the scalability of a single NAS array.
As the number of managed NAS systems increases, the amount of management also increases. Unfortunately, the correlation is not linear. Management tasks increase at a far greater rate than the number of NAS systems under management because the administrator must constantly balance loads on each system as well as migrate data between systems. The management load can be eased with a separate global name space (GNS) product (also called file virtualization) such as those offered by F5, Attune, AutoVirt, Brocade a...
To continue reading for free, register below or login
To read more you must become a member of SearchStorage.com
nd EMC. But implementing GNS still requires a thorough knowledge and understanding of the scalability limits for each NAS system.
There are three primary interrelated aspects of a NAS system's scalability. They include the pragmatic maximum number of files that can be managed per file system and per NAS system (not the same thing) before system performance degrades; the maximum capacity (raw) and usable capacity (amount of storage available after formatting, RAID, data protection, etc.); and the maximum sustainable performance measured in IOPS and throughput. It's quite likely that the limitations related to the number of files under management will be met far sooner than the performance or capacity limits. And performance limitations are more likely to be reached well before the capacity limitations.
Don't make the mistake of just planning for a NAS configuration to meet short-term requirements. If you don't also considered projected requirements, youo're likely to end up spending more and adding to your management woes. Knowing the system's limitations makes it easier to align the right system to the organization's projected growth. It also enables capacity planning to determine when another or more capable system will likely be required. Planning will also help avoid application outages caused when a storage system's limitations are approached or met.
Maxed managed file limits
Managed file limitations are critically important to NAS systems. Few organizations ever ask this question because it is unimportant for SAN storage or DAS. Some NAS systems are known to degrade rapidly or come to a complete halt when the maximum number of managed files is approached or exceeded.
Determining file limitations can be difficult. It depends a lot on file size, file system size and the number of file systems. The NAS system vendor should be able to provide some metrics. Another issue to consider is if you plan to use the NAS system in a multiprotocol configuration to accommodate block storage as well. Many NAS systems provide a Fibre Channel or iSCSI SAN block storage interface. Most of the systems convert those storage blocks (typically 512 bytes each) into a separate files. This will rapidly increase the number of managed files which, in turn, increases the probability of hitting that NAS system's limits.
IOPS and throughput limits
For NAS systems that support NFS and/or CIFS, performance is measured in IOPS and/or overall throughput. Knowing the performance limitations allows alignment of current and future requirements, and helps ensure that the NAS system will perform adequately not only with your current workload but also with anticipate workloads. The key performance limitation metrics are based on the maximum ratings per NAS system and maximums per 1Gbps Ethernet and 10Gbps Ethernet ports.
Capacity limits
Capacity limitations can be a tad trickier than one might expect. The key is usable, not raw capacity. For all NAS systems, vendors list the raw capacity limits and not the capacity that will ultimately be usable. While an exact definition of usable capacity is elusive, the pragmatic definition that should be used is the amount of usable storage available for data after the RAID sets are defined and snapshots are set up. This will vary by drive capacity, and there are tradeoffs between drive capacity and performance. The higher capacity drives spin at a much lower rate than the lower capacity drives which noticeably reduces performance. The higher capacity drives often have lower reliability as well. It's important to make sure the usable capacity limits with the drives selected will meet your company's current and projected requirements.
It takes a little homework and planning to determine the scalability of a NAS system, but knowing the limitations of a NAS system upfront will greatly simplify its management now and in the future.
About the author: Marc Staimer is the founder and Chief Dragon Slayer (CDS) with the 11-year-old marketing analyst and consulting firm Dragon Slayer Consulting. He is a regular contributor to SearchStorage.com.
