Guest blog by John Webster, Senior Adviser, Evaluator Group
Object is Everywhere
Object storage is ubiquitous. You can find it in any medium-large scale enterprise data center. Its available from all of the major cloud services providers. And you’re probably accessing it when you log on to Facebook, YouTube, Instagram and a host of other social media apps.
In spite of its ubiquity though, its still commonly thought of as secondary storage. Object storage is great for data protection, digital archive and other large scale content repositories, but not as primary storage for critical business applications. However, that perception is changing as users discover new primary storage use cases and vendors bring a growing number of high performance object storage systems to market.
High Performance Systems
As the cost of Flash storage goes down, this type of storage becomes increasingly attractive for object storage systems for two reasons:
- A Flash-enabled object system can deliver performance at scale to support new AI-powered, critical business applications.
- By replacing HDD-based object systems that are commonly refreshed in 3-4 year cycles, Flash-based system longevity is greatly enhanced such that users can expect 6-7 year life cycles, reducing TCO of Flash based systems vs. those using HDD. Examples of these types of high performance object storage platforms include Hitachi’s All-flash G10 targeted at information security (Splunk), IoT and AI applications and NetApp’s StorageGRID which can often be found in healthcare IT settings.
Object’s Growing List of Primary Use Cases
The key to understanding the rise in the number use cases is the fact that object storage can now deliver large scale performance at lower cost relative to the block and file alternatives. This coupled with the increasing ubiquity of the S3 API makes object storage a cross-industry and cross-platform standard. Some examples include:
Object Storage for Healthcare
Healthcare’s potential for exploiting high performance object storage is an instructive case in point. Evaluator Group is familiar with a regional healthcare provider and teaching academy that consolidated copies of medical records data, lab result information, and SAP extracted data to an on-site object-based repository containing 14 million active documents. Data types include PACS images, DICOM radiological studies and patient data files. The organization created a single source for clinical and administrative data, available immediately to qualified users, which improves the quality and delivery of patient care
Object Storage for IoT
The Internet of Things (IoT) has provided the impetus for a growing list of new use object data cases that combine the need for real time processing with long-term retention requirements. Autonomous vehicles have the potential to produce what is a massive amount of data relative to our current reference point – on the order of between 5 and 20 TB of data per vehicle, per day. But beyond the shear volume is the complexity of different data types from sensors and cameras. This is coupled with the need to store data that is both immediately available for real time analysis, but also retained indefinitely for legal purposes. High performance object storage is particularly well suited to this application.
Object storage has traditionally been an unlikely use case for transactional applications. That is changing with the advent of software-defined object platforms designed for high performance. An example is MinIO, a software-defined, high performance, distributed object storage system that functions as a cloud-native object server. It will support and is being used for MySQL, PostgreSQL and Teradata applications. MinIO adopted S3 compatibility early on and was the first to extend it to support S3 Select. More than 750 organizations now use MinIO’s S3 Gateway.
It is often said that data has considerable gravity – hence the need to move applications to the data as opposed to the other way around. Object is proving to be ideal for a new wave of applications that are enabled by storage that, at once, offers high performance at scale but at lower cost. It is both transformational and transformable. In my next blog I will consider object storage in another context – a bridge between on-premises data centers and the cloud.