Unstructured data is becoming a big headache for many organizations. By some estimates, this type of data (user files, emails, PDFs, images, videos, etc.) now accounts for up to 90% of all new data growth. While most of this information quickly becomes stale shortly after its creation, many organizations can no longer just simply purge these data stores to shrink their storage footprint. Internal corporate data governance, regulatory compliance mandates and new business intelligence initiatives are making it a necessity to retain unstructured data repositories for longer periods of time. Object storage technology offers a way out of this unstructured data conundrum.
For many years, networked attached storage (NAS) systems have been utilized by businesses of all sizes to store unstructured data. They offered a simple way to add storage capacity over existing Ethernet networks. The challenge was that these “scale-up” systems typically only supported a single controller head to manage all the storage configured behind it. Once the total disk capacity was reached or the I/O capability of the head was exhausted, it would require either an upgrade to a more powerful controller head or the acquisition of a whole new unit. As data grew, these systems would quickly propagate throughout the data center; increasing costs from both an operational and capital expense standpoint.
To address the scaling limitations of scale-up NAS platforms, a new type of “scale-out” NAS architecture was introduced to the market. Scale-out NAS systems deliver all of the same attributes of traditional NAS with one key differentiator – they can scale a single file system out into the multi-PB range. Instead of containing all the compute, memory and storage I/O into a single controller head, scale-out NAS systems leverage a multi-nodal architecture whereby each node has self-contained disk, processors, memory and networking resources. This addresses issues around NAS sprawl and allows infrastructure planners to scale storage and performance out linearly as nodes are added to the system.
While scale-out NAS systems can be effectively leveraged to handle the enormous volumes of data generated into today’s IT enterprise environments, they are not entirely cost effective. In many environments where “cheap and deep” storage is the order of the day, scale-out NAS introduces an additional cost penalty by requiring the coupling of expensive CPU, memory and networking resources each time disk capacity is added to the system. Furthermore, scale-out NAS platforms typically protect data on disk via RAID-6, which adds significant cost overhead as the environment scales out.
Object storage systems, on the other hand, offer much of the same access and deployment flexibility as traditional NAS and scale-out NAS systems; only they enable the use of commodity disk to build-out very large, multi-PB disk storage repositories. Many object storage systems also use a highly disk efficient technology called “erasure coding” to protect data. Some can even automatically expire and purge data based on a set of pre-defined policies. All of these capabilities help to greatly lower the total cost of ownership (TCO) associated with managing unstructured data.
In addition, existing business applications can be coded through an API (application programming interface) to interface directly with the object storage repository for greater application “data awareness”. Some object storage frameworks provide native support of standard object storage protocols like REST while also supporting de-facto cloud protocols like Amazon’s S3 interface. This enables organizations to build-out private cloud environments while maintaining the flexibility to migrate data into public cloud infrastructure.
Another interesting attribute of some cloud storage frameworks, like Cloudian’s, is that enterprise environments or MSPs can use the chargeback facility within the Cloudian software to meter and bill usage based on consumption. In addition, Cloudian is compatible with cloud computing platforms like CloudStack and OpenStack, making integration efforts simple and effective.
In summary, there are still instances where scale-out NAS systems fulfill an essential function; like scientific computing environments where large quantities of data have to be processed extremely rapidly. Likewise, object storage is a natural fit for managing very large, growing unstructured data stores. In short, some environments could benefit by applying the right business use case for each technology.
To learn more about the difference between scale-out NAS and object storage systems, please fill-out the form below for our latest white paper: “Scale-Out NAS vs. Object Storage”, and we will email you a copy of the white paper.