As unstructured data has grown exponentially, requests for file metadata—information such as file attributes and access privileges—have also skyrocketed to become a huge bottleneck for application performance.
Metadata requests are the largest part of file system latency, making up 90% or more of all requests to a storage system or cloud. Performance-intensive apps from AI to autonomous vehicles generate massive amounts of metadata requests.
Read /mnt/homedirs/Bob/test/debug.txt
The InfiniteIO File Metadata Engine (FIME) architecture, powered by deep packet inspection, separates metadata processing from file/O. IME learns file metadata and responds from DRAM at rates impossible to achieve with a file system.
InfiniteIO responds to file metadata requests directly from the network, slashing latency from seconds to less than 100 microseconds. Nothing is faster, not flash, not NVMe.
As simple to install as a network switch – deploys in minutes with no downtime or workflow changes.
With an intelligent metadata management approach, InfiniteIO avoids unnecessary downtime and complexity. You install InfiniteIO like a network switch with no downtime or disruption. Simply drop InfiniteIO Application Accelerator or Hybrid Cloud Tiering products into the network and immediately benefit from faster application performance and the ability to identify and transparently tier inactive files from primary storage to lower-cost NAS or hybrid cloud infrastructure. You can avoid changes to existing applications, storage and clients. The only change is faster performance and bigger cost savings.
Up to 50X Performance Improvement
Up to 80% Storage Cost Savings
The InfiniteIO Application Accelerator immediately reduces latency to dramatically improve application performance. You can reduce latency response time from seconds to microseconds to power applications such as AI/ML, DevOps, HPC, VFX and simulations.
Deliver the fastest metadata performance to improve productivity and reduce app run time
Increase performance up to 50X by responding to metadata requests directly from the network at memory speeds
Add high-performance data tiering with a simple software upgrade. You can quickly identify cold files and automatically place them in any lower-cost NAS or object storage to reduce capex and opex—and access all files with no changes.
Saves storage costsup to 80% by finding and migrating cold data to any cloud or lower-cost NAS
Support billions of files and off-load millions of metadata operations per second without VMs, stubs, symbolic links, client modifications or workflow changes