– Mark Goros, the CEO and Founder of Caringo Inc., says:
The adoption of cloud storage technology for a broad range of consumer and business applications is transforming the storage landscape by transitioning away from traditional disk arrays to object-based storage systems that have the scalability, availability, resiliency and accessibility to enable cloud-scale storage and instant access.
A recent IDC report predicts that the market for File- and Object-Based Storage (FOBS) will experience an annual growth rate of 24.5% through 2017, reaching $38 billion. “Increased versatility will result in more diverse use cases for FOBS,” said IDC.
Software based, object storage is not saddled with the cost, complexity and vendor lock-in of legacy storage arrays or the scalability limitations of traditional file system storage. But all object storage systems are not created equal. Here are six “must haves” that are required to create best-in-class object storage solutions.
1. No Single Point of Failure:
The most efficient object storage systems are built on a symmetric architecture where all nodes run the same code, resulting in high availability and unprecedented scalability, eliminating any single point of failure.
Why this matters to you: When you hear management node, controller node, or database this means more management and the addition of single points of failure that can critically impact performance, stability and fault tolerance. In highly available object storage solutions all nodes do the same thing so that if one fails the others can immediately remedy the issue. This also eliminates the need for specialized hardware that needs to be physically shipped if an issue is discovered.
2. Flexible Data Protection on a Per-Object Basis:
Data protection flexibility is critical as no single data protection scheme can be optimized for every use case. Object storage systems need both replication and erasure coding, as well as the ability to move between them, all available in the same cluster to ensure comprehensive, efficient data protection.
Why this matters to you: One size fits all just does not work in real life. Different use cases require different combinations of replication and erasure coding. Object solutions that constrain the transition from one protection scheme to the other or lock the protection scheme to specific hardware ultimately hinder growth and your ability to optimize resources. Support for both protection schemes on the same server means you can optimize for access, data protection and resource utilization system wide – without constraint.
3. Support for Large and Small Files:
Object stores must be designed with the versatility and flexibility to handle a broad range of applications and workloads without performance impact, equally adept at storing and accessing billions of small files, documents and emails or very large files like high-definition videos.
Why this matters to you: This is primarily about performance from an access perspective. The variation in file sizes will continue. While compression algorithms get more efficient in making files smaller, technological advancements will continue to add to the complexity and depth of some file types resulting in larger files. An object storage solution that ensures rapid access and efficient storage, regardless of file size or object count will increase the number of use cases reducing the number of point solutions you need to purchase.
4. Granular, Automated Scalability:
Best-in-class object stores should support highly flexible scalability, spanning the addition of a single disk all the way up to multiple nodes to extend the capacity or performance of the solution.
Why this matters to you: Granular scalability lets you scale as you grow and eliminates the need to over purchase hardware because of the technical limitations of the storage solution.
5. Continuous Integrity Checks and Fast Volume Recovery:
Best of breed solutions continuously check content integrity from a protection scheme and content perspective. If a bad disk is discovered, recovery should be distributed with the rate of repair accelerating as the storage solution grows.
Why this matters to you: Content you store should always be available. Some object solutions only check data integrity on reads – the worst time to ensure data integrity. Others rely on specialized nodes to identify and repair issues which limit scale.
6. Instant Content Lookup and Retrieval:
Best of breed solutions allow queries against the object store based on object attributes or customizable metadata “tags” stored with the object.
Why this matters to you: As the amount of content grows from millions to billions of objects and management resources change (hardware migration and employee turnover) efficient content lookup and retrieval becomes a challenge. Some object solutions store metadata in a database, which introduces an additional layer of complexity between content requests and content delivery – a textbook bottleneck. Databases also become unwieldy with size and require investment in specialized management resources. By storing metadata with the object, content is self-contained and security, authentication and all identifying information is always available regardless of application, employee turnover, technological obsolescence or even time.
To learn about how to evaluate cloud storage options, identify the commonalities and differences among solutions and get a cheat sheet to assist in your evaluation, click here for an on-demand webinar hosted by Caringo.
Caringo provides pure object storage software that combines ease of management, intelligent automation and elastic data protection transforming commodity servers into massively scalable, fault tolerant storage that preserves your data in addition to your resources and time. Caringo gives you control over the volume, velocity and variability of unstructured information associated with cloud storage, big data and active archives.