AFM gateway node

On AFM cache clusters, AFM uses gateway nodes in order to connect to the home system. Each AFM-enabled fileset uses a designated primary gateway node in order to connect to home and fail over to other gateway nodes as necessary.

AFM gateway nodes might generate a large amount of network traffic between themselves and the home system to fetch and synchronize files. The bandwidth and latency on this network can directly impact file operations on AFM-enabled filesets. In order to ensure best performance and cluster stability, it is best to have AFM traffic use a different physical adapter than the IBM Storage Scale cluster network. It is best to use designated gateway nodes that are not used for other application workloads. AFM uses additional node memory and cache entries on gateway nodes, so applications that run on these nodes compete for cache usage, which slows both the application and AFM operations. AFM gateway nodes are required to be separate from the servers that provide storage for IBM Storage Scale RAID.