SwiftStack’s distributed shared-nothing architecture was designed from its inception to scale linearly and flexibly, which makes it the perfect answer to the ever-more-challenging requirements of today’s modern applications. Capacity and throughput bandwidth can be scaled independently or in parallel and can be added without disruption. Industry-standard servers can be mixed and matched to optimize hardware costs, and with SwiftStack’s software tying it all together, you have your own cloud storage that scales to hundreds of GBytes/sec and Exabytes of capacity.
When you need to minimize latency and maximize IOPS, flash is king. In fact, some SwiftStack deployments use flash for specific functions. And flash can even provide high throughput rates, but if you need to grow to petabytes of capacity, flash presents two major challenges: First, administration and organization of data across multiple silos of flash is a burden, and all that flash is going to cost more than you want to pay.
With SwiftStack, production customers regularly achieve dozens of GBytes/sec (even over 100 GBytes/sec!) in throughput and manage dozens of petabytes in just a few hours each week. No matter how large the deployment, the data is always in a single namespace, and the cost is a mere fraction of comparable flash-based capacity. And for those who do need a latency-optimized tier of storage for portions of a workflow, a combined strategy can provide the best of both worlds: Use SwiftStack as your scalable data platform, and use flash for your short-term active workspace.
| SwiftStack | Flash | |
| Massive Throughput | ||
| Highest IOPS / Lowest Latency | X | |
| Single Namespace, No Silos | X | |
| Linear Scalability | X | |
| Low-Effort Administration | X | |
| Low Cost per Capacity | X |
No other on-premises object storage was architected to scale like SwiftStack. From its inception, SwiftStack’s storage engine was purpose-built to be just that—an infinitely scalable, distributed, shared-nothing, modern storage platform. While some vendors repurpose other technologies to apply them to data storage, every line of code in SwiftStack software is focused on efficiently and effectively managing data, so it’s no surprise that customers and objective third parties repeatedly demonstrate that SwiftStack outperforms the competition.
The kind of ultra-scale performance SwiftStack provides isn’t just helpful in theory; it makes a practical difference in critical real-world business. And while SwiftStack’s scalable capacity and low cost point apply to hundreds of business workflows, the number that require massive throughput across massive data sets is growing quickly.
A lot of data from the survey cars has to be moved and stored quickly.
The “working set” for training data measures into the 10s of PB—which makes public-cloud egress costs prohibitive.
Thousands of GPUs need to be fed subsets of that data simultaneously to be leveraged effectively—which exceeds what alternative storage architectures can deliver.
Only SwiftStack’s shared-nothing distributed architecture was designed from its inception to scale linearly to infinite levels of throughput, concurrency, and capacity, which means only SwiftStack can ingest, retain, and feed training data fast enough to fully utilize thousands of parallel GPUs working to produce today’s most intelligent DNNs and ML algorithms.
Learn more at swiftstack.com/ai.
Many media workflows ingest, edit, distribute, and even archive on the same NAS storage tier, which means there is often capacity pressure when new content arrives. This stalls productive work while old content is moved or deleted.
High-speed transfer tools like Aspera can only move data as fast as the storage can read or write.
Only SwiftStack’s architecture scales linearly to infinite levels of throughput, concurrency, and capacity, which means content can be ingested, distributed, and archived directly on SwiftStack and moved quickly to or from short-term flash storage when needed.
This faster loading and unloading of data means less time spent waiting for the next step in the workflow.
Learn more at swiftstack.com/media.
As existing large-capacity use cases grow to require high throughput and new extreme-throughput use cases emerge, it is necessary to depend on a storage platform that is designed to scale. SwiftStack’s distributed shared-nothing architecture was designed from its inception to scale linearly and flexibly, which makes it the perfect answer to the ever-more-challenging requirements of today’s modern applications.
And not only is SwiftStack’s software the most scalable in the industry, but SwiftStack is also leading the industry to rethink how performance is understood in the world of data storage and management. (Hint: New metrics are important, and new benchmarking tools may be required.)
If you would like to know more about SwiftStack’s experience or how performance affects today’s large-scale workloads, please contact us!
