Object Storage and its Role in Long-Term Storage Today

Reading time for this article .

In my last blog, I discussed types of “data movers” including backup, archive, HSM and migration. These are ways to move data, but where are you moving the data to? Disk? Tape? Cloud? A combination of those? The significance of “data mover” applications to an organization’s workflow is closely tied to other elements of their data storage ecosystem, namely storage targets.

Over time, storage targets have evolved from repositories completely “unaware” of the data they hold to hyper-intelligent data management platforms. On one end of the spectrum, more typically found in tier-1 storage, are Storage Area Networks (SANs). These repositories, which are unaware of the data they hold, offer up blocks of storage which a server’s file system or a database administrator must configure into usable storage. With the advent of Network Attached Storage (NAS), storage devices use their own processing power and file system to lay data across blocks – presenting the storage as a folder of files which can be accessed by multiple servers, and even servers with varying operating systems and/or file systems, across the network. SAN and NAS are commonly found in tier-1 data storage, although they can be used in tier-2 storage as well.

Object storage – commonly used in cloud storage – is the next step in moving “storage intelligence” closer to the storage target, and has revolutionized modern storage.

Progression of data storage

Earlier this month, we presented an introduction to object storage in a blog called, “Mind the Tipping Point: Object versus File Storage”. One of its key characteristics, the unique object ID, deserves further examination. When a piece of information (usually in the form of a file) is moved to object storage, it has to be converted into an object through a “gateway” or “engine,” where each piece of data receives a unique object ID. Unlike specific blocks mapped out by SAN or hierarchical files in NAS, this unique object ID is not tied to a physical location. This specific attribute has had a significant impact on data movement, and that’s where the magic comes in. The analogy commonly used in describing object storage is valet parking. You go out to dinner, give your car to the parking attendant, and they give you a parking receipt. Maybe it’s a nice car, so they park it right in front of the restaurant.  But while you’re eating, an even nicer car pulls up to park – your car gets moved to parking across the street and the new car is parked in its place out front. As the night progresses, the spaces in front of the restaurant and the parking across the street are filled. Your car has been sitting there for hours so they move it down the block to underground parking. When you finally pick up your car, you give your parking receipt to valet parking, and they give you your vehicle back. Intelligent object storage platforms can automatically move data based on storage policies which reflect the “value” of the storage tier in which it resides and the amount of total time since it was last used. When a user or application needs to retrieve a specific object, the unique object ID allows them to seamlessly retrieve the file, none the wiser as to which storage tier it was last located in.

As mentioned above, object-based storage is commonly used in cloud storage. When users back up their data to the cloud, they don’t know (exactly) where their data is. Cloud users are typically unaware of whether their data is stored on disk or tape, or online or offline. Cloud providers offer various service level agreements that agree to return data in a specific amount of time, and users may be able to specify that they want it stored in a separate geographic region for disaster recovery (DR) purposes, but that’s about the extent of it. As data-driven organizations consider cloud storage, they should keep in mind that the infrastructure of the cloud (http and object storage) is a large part of what makes it so appealing. When implemented locally in data centers, this infrastructure can truly modernize on-premise, long-term data storage.   Spectra’s BlackPearl® Converged Storage System incorporates a feature-rich object storage engine. The innovative solution combines multiple storage targets into a simple and affordable self-managing, cloud-enabled object-based platform that provides organizations with the openness, scalability, efficiency and control they demand to easily grow and adapt to changing business models.

BlackPearl can retrieve data from the fastest and/or most affordable storage when multiple copies are kept, checksum the data before returning it – and if it’s no longer good – find another copy to return while also replacing the failed copy. For environments with tape libraries as a storage target, BlackPearl can compact tapes as certain data sets age off to maximize free space and even migrate data from older versions of LTO tape to newer versions of LTO tape without disrupting operations. While there are “data mover” software packages that can deploy object storage by writing data to the cloud, they don’t always play well with other forms of storage. By allowing these data movers to send data to BlackPearl, configurable data policies can send copies to the cloud, archive disk, tape or replicate the entire setup at another site – all in object format. This “valet parking” approach allows users to select the appropriate storage target for access and cost based on the data’s business value. As the value of that data changes over time, BlackPearl can move the data to different storage areas accordingly. Organizations can and should use object storage in their data centers. As more applications adopt object-based interfaces, object storage will become the de facto standard for data movement and management of long-term storage.