Home > Storage > PowerMax and VMAX > Storage Admin > Reliability, Availability, and Serviceability on PowerMax 2500 and 8500 Arrays > Global access and BEaaS
In PowerMax 2500 and 8500 arrays, media enclosures are accessible globally using a central fabric. This gives ‘any to any’ access, which means that all IOMs drive IO to all the disks, which improves backend load balance and improves resiliency.
BEaas (Backend as a Service) picks the IO path based on IO complexity and load balancing.
Dynamic media enclosures are highly resilient. Each unit contains:
Each interface module has access to all drives in the enclosure. In terms of fault tolerance and redundancy you could lose an entire power zone, and a PSU from the other zone, and still have full access to all data with one interface module and all drives powered.
Optimum reliability is implemented using a Markov model in a Flex RAID implementation, which is a new RAID distribution model. The backend capacity is distributed on a set of drives and TDATs are configured based on the RAID type.
This model improves rebuild efficiency and offers flexible capacity expansion while meeting the availability and reliability numbers of traditional RAID.
With PowerMax 2500 and 8500 and Flex RAID, the sparing model changes. In place of spare drives, there are spare hypers. The spares are provisioned at the Disk Group level, so that all drives that belong to that Disk Group will share the spare capacity. This reduces the spare requirement at the system level and speed for rebuild due to any-any drive access.
Direct Member Sparing (DMS), which is used to back up data from a failed drive to spares while host IOs are running, is also supported on PowerMax 2500 and 8500 platforms. When a drive fails, all the hypers on the failed drive will be rebuilt to spare hypers on different drives, in the same Disk Group.