Here are a few recommendations. 

1). Provide min, max heap size while running the ASMM.jar file. ASMM will run in a Jetty server (JVM) so you need to provide the heap size. Depending upon the load you need to provide these options when running.

2). Remote client connections to ASMM is recommended. In case you want to connect a regular client, make sure to run ASMM on the same host where other agents are running or keep the ASMM host in the same subnet where other cluster agents are running.

3). Do not run any large queries without a filter condition on it or a non-indexed column.

4). In case you want to run large queries, ASMM is not the right place. You can run the queries through the JAVA API or a BW Client. 

5). Do not load the ASMM tool with too many users. Do not expose the end point URL to users.

6). Default "Maximum browser entries return: 3000 entries " in server.prop file. You can reduce this to 1000.

7).  Provide the metaspace and discovery URL in the server.propriety file so that from  the UI you can provide only the metaspace name to connect to the cluster. The remaining details (discovery URL,listen URL) are picked up from the server.propriety. With this approach you never share the discovery URL with the user.