InfiniteIO has announced Application Accelerator, which delivers performance improvements for critical applications by processing file metadata independently from on-premises storage or cloud systems. The platform provides organizations across industries the lowest possible latency for their mission-critical applications, such as AI/machine learning, HPC and genomics, while minimizing disruption to IT teams.
The Application Accelerator allows organizations to implement innovation, with minimal to no disruption to existing IT operations. It helps to offload and respond to metadata requests at the speed of DRAM, which improves performance up to 50 percent for workloads such as machine learning training models; and delivers over 3 million metadata operations per second, doubling the performance of the previous-generation product.
The platform deploys without requiring changes to existing applications or storage systems; supports enterprise growth by scaling to 1 billion files in a single node or 2 billion files in a clustered configuration; enhances the performance of all NAS systems, even all-flash and NVMe arrays; and can be upgraded with high-performance data tiering capabilities.
“Bandwidth and I/O challenges have been largely overcome, yet reducing latency remains a significant barrier to improving application performance,” said Henry Baltazar, vice president of research at 451 Research. “Metadata requests are a large part of file system latency, making up the vast majority of requests to a storage system or cloud. InfiniteIO’s approach to abstracting metadata from file data offers IT managers a non-disruptive way to immediately accelerate application performance.”
As unstructured data has grown exponentially, requests for file metadata—information such as file attributes and access privileges—have also skyrocketed to become a major bottleneck for application performance. InfiniteIO’s latest release, built on the InfiniteIO Metadata Engine (IME) architecture, responds to file metadata requests directly from the network instead of the network-attached storage (NAS) or cloud storage system.
InfiniteIO’s metadata abstraction can reduce latency from seconds to microseconds for all files in a hybrid cloud environment. This results in faster access to data and speeds up application performance.
The IME collects metadata for the files from storage systems and stores it in a memory-based metadata map. All metadata is always hot, unlike a cache, for the fastest response. After a one-time scan of the file system, InfiniteIO’s deep-packet inspection technology keeps metadata current in real time. InfiniteIO improves application performance regardless of the underlying NAS vendor or storage technology, including all-flash or NVMe arrays, by serving metadata requests out of memory as fast as 40 microseconds and up to 3.2 million metadata operations per second.
“Reducing latency is the last frontier in improving application performance. The tech industry has been focused on making incremental performance improvements with faster storage and file systems, when the biggest opportunity is in removing the file system latency created by processing metadata requests,” said Mark Cree, CEO of InfiniteIO. “Separating metadata processing from file I/O significantly decreases application latency, which translates into reduced product development cycles and greater worker productivity.”
InfiniteIO also released new software features that continue to simplify and accelerate tiering of cold data from primary NAS systems to lower-cost cloud storage, such as the ability to scan 1 billion files in a day and a new API for cloud usage charge-back. The IME architecture never recalls metadata back from the cloud, increasing performance and avoiding cloud egress charges.
Robust policies automatically tier files so that even rarely accessed information is available on-demand, without disruption or performance compromises.
The InfiniteIO Application Accelerator is available immediately from InfiniteIO’s growing network of international resellers and solution providers. Existing customers with current maintenance agreements can add the new high-performance data tiering features with non-disruptive online upgrades.