Supported back end storage is restricted to some storage from major players (HPE, EMC, IBM), so there is room for improvement in supporting more storage solutions.

According to my experience, for certain HPC environments, we encounter certain Memory/CPU limitations when we have a large number of IOPS. This is why, in order to support such loads, several cabs are required in parallel, complicating the configuration and administration.

Another handicap is the management of the LACP that, when we are talking about environments with less than 20 calculation nodes, may not balance correctly despite setting the RR option.