Hi, are there any recommendations/ guidelines for the maximum number of Vision/perspective clients per gateway. I know that theoretically the number of clients is “unlimited” and is dependent on the hardware resources of the gateway, so are there any tests or benchmarks that were done to verify the practical limits. I have a project with a design capacity of 160 concurrent clients.
Another question is it possible to have one project that will be synchronized between 2 gateways (cluster setup with load balancing), if I am not mistaken the redundancy does not seem to offer a load balancing feature (only a backup if the primary fails: one gateway is active at a given time).
Thanks in advance.
This is a hard thing to provide accurate guidance on because it is so strongly dependent on the design and behavior of the project being run by these clients. 160 is a reasonable number assuming a fairly light-weight project design (few bindings / bindings that subscribe to data that changes infrequently / slow polling rates for queries / queries that don’t bring back large datasets).
Yes - it is definitely possible to synchronize a project between two (or more!) gateways and put them behind a load-balancer. You’re right, our redundancy functionality is not about load balancing - it is squarely focused on reliability. To sync a project to multiple load balanced gateways, most customers use the EAM module and gateway network connections. Make sure you use a load balancer that supports sticky sessions.