Cloud object storage and scale-out storage are both good options for storing large quantities of data. While both...
can be beneficial from a management perspective, object storage is on the rise as an alternative to scale-out storage for its improved capabilities in several areas.
Scale and efficiency
Traditional storage architectures, whether scale-up or scale-out, use a hierarchical structure similar to a file system tree to manage data blocks in the storage system. The information that facilitates this data management-- where it's written, which blocks are associated in files, access information, etc. -- is called metadata, and it grows as the storage system fills up. Like a file system, there's an upper limit on the number of data blocks a storage system can manage, one that's related to the system's ability to process a growing set of metadata.
Instead of the file system-like structure, cloud object storage uses a key-value structure to organize data that creates a simple index of object IDs, (the key) each of which is associated with the location of a discrete data object (the value). Instead of creating metadata that describes a block's location within a multiple-tier hierarchy, an object storage system just stores its ID number. This creates a flat organizational structure that's much more efficient than traditional storage architectures. They generate far less metadata and require fewer metadata processing steps to access data, enabling object storage systems to grow much larger.
Cost at scale
Data protection in traditional storage systems requires making extra copies of files, typically stored in multiple locations. As these systems grow, the cost of creating those extra copies can become prohibitive. Object-based architectures are ideally suited for technologies such as erasure coding and data distribution because they can provide this data protection and resiliency by expanding metadata without creating multiple copies of each file.
Related to the difficulty of storing extra copies of data is the use of RAID to provide data resiliency and ensure availability during component failures. When the drives in a RAID-based system get very large, the rebuild cycle can take days to complete, rendering the system vulnerable to a second or third component failure. Erasure coding cloud object storage architectures eliminates this long rebuild time and supports increased levels of resiliency to meet a company's risk profile.
RAID's place in object-oriented storage
When to implement a cloud object storage platform
Object storage ideal for rampant data growth
Related Q&A from Eric Slack
With increasingly demanding workloads, NAS systems can have a hard time keeping up. Can using NAS with an object storage option relieve some of these... Continue Reading
Erasure coding is an effective tool for correcting errors in data systems, but when used with object storage systems, latency issues can arise. Continue Reading
Eric Slack of Storage Switzerland discusses whether it is necessary to back up data stored in file sync services in this Expert Response. Continue Reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.