AmirusMM™ makes an excellent substitute for hardware reflective memory when

• Cost is a critical concern. Mirror Memory doesn't need special-purpose hardware or cabling infrastructure, so your initial investment is far less.
• Read-performance is important. Mirror Memory uses cacheable main memory, so time-sensitive algorithms can run at full processor speed, without having to get data from the IO bus.
• You need the flexibility to add new nodes without having to rewire the factory, lab or office . Mirror Memory can run over a shared EtherNet, using your existing infrastructure, so extensions are simple and inexpensive.
• Your application design needs foolproof synchronization -- Mirror Memory has a lock API for coordinating access to shared memory, and automated 'CopyUp/CopyDown' protocols to ensure that newly connecting nodes see the same memory contents as existing ring members
• You are developing heterogeneous applications which must run on machines with different backplanes -- Mirror Memory needs only a network adapter and the supporting IP stack in software.
• You need to implement a reflective memory system over a larger geographical area than permitted by hardware reflective memory system constraints.
• You are implementing applications that will eventually run on hardware reflective memory, but will benefit from the flexibility of AmirusMM environment during development..

The performance of Mirror Memory is dependent upon the processing power of the CPU on which it runs. The distribution comes with a variety of monitoring and test utilities to verify the performance of an installation.

Typically, the per-machine bandwidth will be over 30Mbits/second, depending upon ring characteristics.