To say the data world has changed over the last few years is a gross understatement. Computers started out in the finance department doing accounting because you couldn’t hire accountants and clerks fast enough. They quickly spread to other departments to improve the efficiency of their operations, and today they are underlying and driving every internal and external-facing activity in our organizations. Yet even amid today’s technology revolutions such as the cloud, AI and IoT, for most organizations the foundations of the early computer revolution are still around. The challenge is how to accommodate petabyte-scale data and data growth of 50% or even more in a cost effective, manageable way?
This morning Western Digital announced the ActiveScale™ P100 system, an extensible, cloud-scale object storage system to provide a great way to deal with today’s data growth needs. The system smarts required to manage this growth is found in ActiveScale OS 5.0, the latest generation of advanced erasure coding, combined with the new ActiveScale System Management (SM). Also being announced is our cloud-based storage analytics capability known as ActiveScale Cloud Management. Now our customers have the tools to manage the explosion of data, and how we’re helping them succeed with 50% data growth.
Traditional Architectures Struggle in a New Data World
Traditional hierarchical file structures found in block storage for database applications like Oracle, SAP, and DB2 are the foundation of IT architectures. They tend to bog down when things get too big because the hierarchical structure forces data up and down the hierarchy which can impact performance. Traditional systems exist to provide high performance for transaction-based databases and they are expensive. When your average transaction database is about 5TB, things work just fine for that scale. When you move to petabyte-scale storage in a hierarchical structure, things can grind to a crawl.
Traditional IT architecture uses RAID (redundant array of independent (or inexpensive) disks) to address data integrity concerns from disk drive failures. Because of the performance requirements for databases, RAID needed to be pretty fast, and RAID delivers protection and performance for small scale use. As RAID moves into large scale use, problems arise. As disk drives move to 6TB and larger, the amount of time it takes to rebuild that drive can increase the probability of data loss. Nobody likes data loss, so a different approach is needed for large scale data integrity.
A New Era of Object Storage
How can you keep up with 50% data growth rates? For data that doesn’t need super low latency, there is an alternative. Object storage is designed for large-scale implementation, although it doesn’t have the super low latency of RAID. Hierarchical RAID-based storage still works for smaller storage with high performance needs. No one approach will fit all your data. The wrong approach will probably cost more than it has to and can create data integrity problems.
For most non-transactional data, object storage may be the better approach. Object Storage has a flat file system using metadata to keep track of the objects, avoiding the hierarchical structure that gets more cumbersome at petabyte scale. Object storage also uses erasure coding instead of RAID to protect data. Erasure coding doesn’t rebuild a physical drive like RAID, it recovers a logical collection of data called an object. This provides flexibility that enables object storage to scale to petabyte domains with reduced fear of data loss.
Industry-Leading Object Storage Technology
Western Digital object storage takes data integrity to the next level with advanced erasure coding that provides up to 15 nines of data durability (99.9999999999999%) for data integrity, a level of data durability that RAID simply cannot approach effectively.. Our object storage provides advanced erasure coding instead of RAID, a flat file system instead of hierarchy and geo-spread (known as BitSpread®) instead of replication. In addition, you get strong consistency for always-fresh data, which not all object storage solutions provide. If you’re serious about data integrity, the answer is ActiveScale.
Our geo-spread system architecture has availability built-in. A 3-geo configuration can tolerate an entire site becoming unavailable and all your objects are still available, and at a lower cost with simpler management than traditional replicated sites. In fact, we offer one management instance for the entire namespace, so a system spread over 3 geographies will be able to be managed as a single system. If you think system availability is important ActiveScale is the answer.
We’ve listened closely to our customers’ needs and launched many great new management features with the ActiveScale P100. The ActiveScale System Management (SM) offers real-time management & monitoring, guided wizards, real-time system health, one click upgrades, and APIs for automation to help ensure users see greatest efficiencies from their hardware and have transparency into their operations. You can tune the erasure coding in our new ActiveScale P100 to optimize durability or capacity- depending on your application needs and see current status as well overall usage trends. Data integrity is additionally enhanced in our object storage with continuous background verification known as BitDynamics™.
With a base capacity of 720TB raw that can grow to over 19PB, the P100 is an extensible system to help you take advantage of object storage at your scale today, and easily expand as you grow. Some of the key benefits and differentiators of our object storage operating system includes:
- Scalability scale-up and scale-out to meet your growth needs
- Geo-spread system availability that is less expensive than replication
- Rateless Erasure Coding for high resilience, up to 15 nines of data durability
- BitSpread for automatic optimal distribution of data across available storage to avoid hot spots
- BitDynamics continual data integrity verification with over 6,000 equations
- Encryption provides end-to-end protection for sensitive data
- Strong Consistency for always-fresh data
- Tunable adjust encoding to optimize capacity or data durability
ActiveScale P100 object storage can also easily integrate with existing workloads or cloud configurations. For example, you can write NFS/CIFS/SMB files or Swift objects to ActiveScale for archive, backup and repository needs with an optional gateway.
Building the @Scale Data Center
The key to the @Scale data center and to supporting next-gen workloads with massive data growth is to use the right tool for the job. Put your database on traditional fast, hierarchical RAID-based storage. Put your workloads with lower performance requirements on ActiveScale object storage for lower cost, higher resilience, and superior data integrity.
ActiveScale is the right tool for your petabyte-scale data center or cloud, and we’re thrilled to share our latest innovation with you today.
To learn more about the ActiveScale P100 system: http://www.hgst.com/products/systems/activescale-system
You can also schedule a test drive, anywhere around the globe. Contact our team here.
Join Me for a Webinar!
Want to learn more about how to prepare for the future of data? Join me next week as I’ll be discussing how to contend with the explosion of data beyond just capacity management:
How to (Better) Manage 50% Data Growth Nov 22 at 9:00 am PT Register here
Erik is the Senior Director, Product Marketing of Western Digital's Data Center Systems, with 25+ years of experience in high tech storage.