I really like WEKA NeuralMesh's ability to handle my workloads compared to legacy storage systems like Lustre. It's flexible and (so far) reliable, allowing me to do live software updates and work on individual backend servers without taking the filesystem down. The support for multiple protocols is great; it lets me integrate it seamlessly into my end-to-end workflows, like pulling data from Glacier and landing it on the high-speed filesystem for simulation work. WEKA NeuralMesh effectively supports high IOPS and high throughput workloads, which is a major plus for my tasks. Review collected by and hosted on G2.com.
The price is way too high. The need for clients to dedicate CPU cores to the filesystem is difficult to handle with traditional HPC schedulers that allow exact placement of processes on sockets and cores. Review collected by and hosted on G2.com.