As Carol Sliwa discusses in a SearchStorage.com article, “Scale-up vs. scale-out network-attached storage,” data growth has given birth to a number of new storage architectures, one of which
Scale-up NAS systems use an architecture similar to traditional block-based disk arrays. They have a storage controller that connects to the network and provides I/O to users. The controller also connects to disk drives on the back end, typically with Fibre Channel or SAS. The knock against them has been scalability. When the system begins to reach capacity, the controller becomes saturated and performance suffers. For a number of years, the only solution was to buy another NAS system. For large companies, this strategy could mean a data center with multiple silos of storage.
Scale-out NAS provided a way to address this silo problem, by putting disk capacity into modules that combined storage capacity and processing power. When the system needed to scale, modules were added, allowing the processing capability to scale along with the drive count. There were some problems related to connectivity between modules and managing the distribution of files between controllers, but these issues were worked out, and scale-out systems have been a welcome alternative for many IT organizations.
Hardware-based scale-out NAS systems, from companies such as Isilon, were the early entrants in this sector. These products had impressive scale and functionality and were aimed at IT organizations that couldn’t just keep adding more NAS filers to support their needs for storage. As this technology has matured, scale-out solutions have become a common architecture for cloud and “big data” analytics as well.
But scale-out NAS is also appealing to smaller companies with more traditional storage needs. There are software-based solutions available now that install on commodity hardware to provide an economical, highly scalable storage system. These products, from companies such as Caringo and Red Hat Gluster, feature a global namespace that can scale into the petabyte range but also provide an appealing storage solution for midmarket companies.
The scale-up vs. scale-out decision used to be pretty clear-cut. If a customer had the data growth potential to outpace a single NAS filer, it was a good candidate for scale-out storage. But now, with the addition of commodity hardware to its original pay-as-you-grow configuration, scale-out NAS has greater appeal. It can be an attractive solution for companies far below the sweet-spot size of those original scale-out products. And with technologies such as object-based file systems, it can provide a solution that reaches far above it. This is the interesting development that makes scale-out NAS an excellent technology for VARs.
Data growth is a given, and file data growth is outpacing block-based data growth in every segment of the market. Scale-out NAS is becoming an architecture of choice as companies look for cost-effective ways to stay ahead of internal demands for storage.
For VARs it’s a familiar scenario, but one that’s ripe with opportunity. Misinformation typically runs high in areas like this, with so many vendors and such a wide range of solutions. Scale-out storage -- as hardware, as software, as a grid or cluster of nodes or as a combination of the above -- gives customers a lot to get their arms around. VARs that represent multiple vendors have the credibility and the understanding to step in and provide some real value as a trusted storage advisor. Then, when the time comes for proposals, they can leverage their line cards to offer the best solutions.
Eric Slack is a senior analyst with Storage Switzerland.
This was first published in December 2011