Enterprises should standardize on object storage for almost all (95 percent of their unstructured data and limit their network attached storage systems (NAS) storage to only the most active data that requires high performance. The payoff to your organization will be dramatic. Here are my three reasons you should consider cloud storage; lower costs, scalable software and data protection.
There are many storage companies that claim their systems have the ability to scale to almost limitless proportions. But at what cost and at how much complexity? The hardware behind scale-out architectures is common. Use servers to act as storage nodes that are aggregated into a storage cluster. That cluster presents a single storage pool that data can be written to.
Where object storage differs is what hardware it can use to fulfill that need and how far the software itself can scale. In terms of hardware, “heavier” file systems common in NAS storage require much more powerful (and expensive) servers to act as nodes. Object storage software tends to be lighter weight enabling the use of true commodity hardware. The organization does want to be careful how far into this process they want to delve. Most organizations should look for an object storage vendor to provide the object storage software and hardware turnkey. This allows them to benefit from the cost advantages of commodity hardware without them having to become system integrator in their own right.
Truly Scalable Software
The second aspect of software is to what extent can the software itself scale? While many modern NAS file systems claim to scale to billions of files, there tend to be limitations into how many files can be contained per volume, per directory or subdirectory. There may also be limits on the number of directories that the NAS file system can contain. There is also the issue of human scale, just how many directories and subdirectories to you expect the user to navigate?
Finally there is the very real reality that with the ability to support one billion files, the NAS system may not be able to support all the data going to it. The Internet of Things (IoT), with every device generating data that needs storing and to be analyzed, a billion files come very quickly.
Object storage has no such limitations. Because of its flat structure, it avoids the concern over navigating and managing various volume, directory and folder structures. In terms of object count it quite literally can support trillions and trillions of object.
Object storage systems have a variety of data protection techniques available to them. NAS systems are generally limited to either RAID 5 or 6. IT professionals should look for solutions that support both replication and erasure coding. Replication allows for a user-defined number of copies of data to store on separate nodes but at the expense of capacity. Erasure coding is a parity-based scheme similar to raid, but unlike raid is data aware, enabling rapid rebuilds from a drive failure.
A common request of unstructured data storage systems is to keep data for a specific period of time or forever. Applications can leverage object storage’s rich metadata to encode specific expiration dates on files. It can also leverage the unique ID discussed in chapter 1 to verify data integrity, as the checksum should always generate the same ID.
Object storage brings many advantages to the enterprise. Low cost, scalability and data protection are just the top three. Another attribute of object storage is that leveraging the technology is easy. If the object storage system provides support for legacy file systems or protocols, enterprise can move to the object store today with no change to applications. Then as time and ability allows, they can update their software to take full advantage of the technology.
Sponsored by Dell EMC®
Sponsored By Dell EMC®’s whose 3rd generation object-based offering, Elastic Cloud Storage (ECS™), delivers all three reasons outlined above in a single, cloud-scale storage platform. ECS leverages a software-defined architecture with commodity infrastructure that provides an infinitely scalable, geo-distributed system that features an industry low cost to serve.