Reflective memory is way of sharing data between two separate systems in a deterministic way. The systems that share this data are typically different and independent systems. This method was developed in the 1980s for use within VME systems and is typically used with VXI, real-time operating systems, PXI and PCI standards, as well as a variety of other platforms for which solutions can be created.
How Reflective Memory Works
Devices that use this method are connected through the use of fiber optic cables. This forms a deterministic network that operates in a similar fashion to a dual-ported memory system. The memory network is formed when a system within the network acquires data then writes it into its local RAM.
It behaves like a dual-ported system because when one system acquires the data and writes it to its local memory, the data is automatically broadcasted and written to the memory of the second system. Read operations access the local copy of data which provides the most updated data from the network to the user.
These networks work on real-time meaning that each connected device or computer’s individual copy of the shared data set is a local and updated copy. Networks using reflective memory are designed specifically for data communications that are highly deterministic, meaning if they are given the same input information they will produce the same output information.
Common Applications
Reflective memory is best for serving demanding, real-time applications that require things such as lack of software overhead. There are a variety of benefits to using this type of architecture for design, including:
* Performance – high speed, low latency, data delivery
* Operation – user-friendly with easy implementation
* Versatility – connects to dissimilar bus structures and computers with standard and custom platforms and data sharing can occur over any operating system or processor type
This type of memory can be applied to PCI and PXI standards as well as VME systems and VXI. It can also be a applied to real-time operating systems as well as real-time networking with solutions for a variety of other platforms. This network provides the strictly timed performance that is a necessity for all sorts of industrial control or distributed simulation applications.
