Object storage systems are gaining growth in big data circles, especially in oil and gas, media and entertainment. Storage area networks (SAN) need expanding options without sharing primary storage. Some advantages of an object storage system are the lack of necessity to have a single namespace for cache conservancy; instead the owner can store loosely related blocks of data without a metadatabase. The data is based on its own standard, not the physical location of the storage. This allows the data to scale based on the rule or standard, instead of being based on the storage location.
Big data environment companies want node-centralized storage, with big data analytics and archiving opportunities. One of the best advantages of the object storage outside of the cloud is the freedom to deposit material from multiple geographic locations with multi-tenancy. The Albert Einstein College of Medicine of Yeshiva University, based in New York, is using a Web Object Scaler (WOS) for the repository of biomedical research. The convenience of holding a large number of images, the ease of use for the multiple researchers, and the similarity to the Amazon cloud storage system was the winning combination for the research teams.
DataDirect Networks Inc. reports several customers using 10 PB or more with the WOS systems, including one client that stated he has more than 70 million objects in his data management system. Amazon is again credited for the widespread openness to using a virtual data storage system, instead of an attached separate hard drive or a multi-database that requires location proximity. OneBlox appliance, created by Exablock, offers a product with a file-based interface that complements SANs and is also used as a backup target for those organizations with a quickly growing storage need. The primary advantage for this product is the integrative function of VDI with a high-performance SAN for those who want storage for photos, directories and app data but don’t want to use up or clutter the SAN.
There are other options when object based storage isn’t the right fit, such as moving the applications based on block storage to an applications that closely mirrors the application, for example Elastic Compute Cloud (EC2) or Elastic Block Storage (EBS), where storage is created in EBS and then attached to EC2. This is less work than rewriting the applications for object storage but more work than just using cloud storage. A second option is to use a cloud storage gateway approach using an on-site local primary storage connected to storage in a public cloud storage. This requires hardware investments but give an on-ramp to cloud storage, that includes the benefit of using the existing on-location storage for proximity. The last option is to investigate third-party products that provide storage for apps using an external cloud. Red Hat, for example, with their product Red Hat Storage Server for AWS, offer solutions that let you define a private storage array similar to Tier 1 storage, using AWS storage. These selections are most advantageous for companies already in relationship with these third-party dealers.