EnSight 2024 R2 has the ability to use multiple EnSight Servers to process a dataset in an efficient distributed manner by using what's called Server-of-Servers (SOS for short). See Use Server of Servers for details. The SOS process acts as a router between the EnSight Client and one or more EnSight Servers. It broadcasts commands from the Client to the multiple Servers and gathers and aggregates response data from the Servers to send back to the Client. If the dataset size is sufficiently large, a single SOS may become a bottleneck. In such situations, multiple SOS processes coordinated by a parent, or root level, SOS may be warranted to increase scalability. EnSight 2024 R2 has the ability to use multiple SOS processes to distribute the load, and adds a single Root Level SOS (RLSOS) that communicates between the EnSight Client and multiple SOS processes. Each of the additional SOS processes in turn communicates with one or more Servers (see Figure 1). Root Level SOS (RLSOS) enables EnSight to scale to much larger problem sizes.
Just as the EnSight SOS acts as router for multiple EnSight Servers, the RLSOS acts as a router for multiple EnSight SOSes. In theory, the RLSOS should provide the same degree of scalability for multiple SOSes as a single SOS provides for multiple Servers. It is anticipated, but not yet tested, that the RLSOS could enable EnSight to scale into the tens of thousands of EnSight Servers. To take advantage of this potential, the data set must be sufficiently large to keep all Servers busy and to warrant the overhead of the additional communication. It is anticipated that data sets should be on the order of millions of elements/nodes per Server to obtain reasonable scalable performance.
