WEKA.IO - No Limit, No Compromise Data Access
Your company has invested in the best infrastructure to run your AI and data intensive applications, so don't let data accessibility be the bottleneck to your productivity gains. To maximize your investments and resolve the GPU and CPU starvation problem, you need a shared file system which ensures that data is available to the applications. WekaIO delivers all the bandwidth you need, so your applications never have to wait for data.
WekaIO Delivers Record-Breaking Performance Results
WekaIO’s claim to be the world’s fastest parallel file system are substantiated by independent SPEC SFS 2014 test results. Using off the shelf industry-standard components, WekaIO’s software achieved #1 position across all of the benchmark suites including EDA, software build, database, VDI and video streaming
Dedicated Storage Server
Matrix software is an NVMe-native, resilient file system that runs on commodity servers, delivering the highest bandwidth, lowest latency performance to any InfiniBand or Ethernet enabled GPU or CPU based cluster. Integrated tiering automatically and transparently migrates your cold data to any S3 or Swift compatible private cloud for low cost and limitless scale. A dedicated storage server deployment is ideal when you want to maintain separate storage and compute infrastructure for isolation or large capacity scaling.
Our POSIX compliant file system runs more efficiently than NFS and leverages an innovative, customized protocol to deliver file based semantics. As a result, Matrix exceeds NFS performance in a sharable file system solution.
Matrix software can be deployed alongside other applications on your existing GPU or CPU based server infrastructure presenting a POSIX compliant file namespace. It shares the existing Ethernet or InfiniBand infrastructure and provides ultra low latency access to your data. Integrated tiering automatically and transparently migrates your cold data to any S3 or Swift compatible private cloud for low cost and limitless scale.
Hyperconverged deployments eliminate your external storage footprint and lower operational costs when compared to standalone storage infrastructure. Imagine getting better than local file system performance across your entire GPU or CPU based cluster.
Find more information about Weka.io Hyperconverged here >>