Eliminates the need for a multitude of protocols that add costs and complexity.
We had the opportunity to meet with Jeff Denworth, Co-founder & V.P. of Products at VAST Data as part of the most recent IT Press Tour. VAST was founded in 2016 to help simplify 30 years of storage complexity.
For a number of years clients were happy with all-flash arrays (AFA) deployed on hard-drives that resulted in a proliferation of systems and architecture to achieve their needs on the performance and capacity matrix. Jeff referenced one telco with more than 100 different types of storage. Picking up different storage solutions to meet the ever-evolving performance/capacity balance, companies have been migrating data, refreshing data at different rates, and attempting to integrate a multitude of different systems which has resulted in tremendous inefficiencies.
On-premise storage needs to be as simple as that offered on the public cloud or customers will just buy their own cloud-based resource. Companies in many industries are looking for opportunities to apply artificial intelligence (AI) to the data they are already storing and managing. New machine learning (ML) workloads demand random, broad, and real-time access to data. This has turned the storage pyramid of archive (object, cloud, tape), backup, network-attached storage (NAS), AFA, and RAM upside down.
Flash breaks the performance/capacity tradeoff. Performance is a by-product of capacity. It’s impossible to build a slow all-flash archive. The universal storage era is being driven by tier-5 cost efficiency, exabyte-scale file and business object storage, and all-NVMe performance. All data and apps can now reside on one platform which simplifies the work of the data center. Data is always available in real-time.
Companies used to use different data platforms for different jobs:
Block storage for a very high transaction I/O
File storage for easy data services or large-scale unstructured data
Parallel file systems for high throughput access into a machine
Object storage when you need to go “cheap and deep”
VAST addresses all of these and changes how customers think about storage protocols by integrating three technologies that were not available until 2018: NVMe over fabric, low-cost flash, and 3D XPoint Intel Optane Memory. Together, these provide the scale and ability to use algorithms to reduce costs.
The new type of storage architecture, DASE (disaggregated stored everything), scales out while sharing all data. They do this by eliminating east-west traffic with no cache coherency challenges, no batteries, no rebuilds during server failure, and Docker-based auto-scaling based on the real-time needs of the apps. The global architecture is the foundation for global algorithms with global namespace, global flash translation, global data protection, and global data reduction.
Applications can overrun servers unless they are directed to move to another server. There’s a need to isolate traffic from one application to another. A single storage system consolidates noisy neighbors and islands of storage. Server pools enable multi-tenancy.
Global data protection is 15 to 30-times more cost-efficient than competing approaches. VAST codes accelerate rebuild speed by using new algorithms that get faster with more redundancy data. They have achieved a similar breakthrough in data reduction of 20:1 with backup data, 8:1 with market data, and 3:1 with animation and high-performance computing data.
AI is the most popular use case for flash, and VAST also has popular applications with vertical solutions, secondary storage, and enterprise infrastructure. Ginkgo Bioworks is enjoying the reduced costs of flash while getting near-infinite scale. Zebra has addressed its AI storage challenges by getting better performance than NAS with a simple, scalable appliance that’s easy to deploy and manage. The Department of Health and Human Services has used VAST to move beyond hard drives with software-enabled storage architecture to modernize their science and enable AI-driven research with flash. SquarePoint Capital is using VAST’s universal storage platform to support the intensive computing efforts of hundreds of researchers on a global scale.