diff --git a/docs/configuration-large.md b/docs/configuration-large.md
new file mode 100644
index 0000000000000000000000000000000000000000..5dc3c03e799c34ad700da2106f55baa165e64bc8
--- /dev/null
+++ b/docs/configuration-large.md
@@ -0,0 +1,27 @@
+# Orchestrator configuration in larger environments
+
+If monitoring a large number of servers the backend database can become a bottleneck.   The following comments refer to using MySQL as the orchestrator backend.
+
+Some configuration options allow you to control the throughput.  These settings are:
+- `BufferInstanceWrites`
+- `InstanceWriteBufferSize`
+- `InstanceFlushIntervalMilliseconds`
+- `DiscoveryMaxConcurrency`
+
+Limit the number of concurrent discoveries made by orchestrator using `DiscoveryMaxConcurrency` and ensure that the backend server's `max_connections` setting is high enough to allow orchestrator to make as many connections as it needs.
+
+By setting `BufferInstanceWrites: True` in orchestrator when a poll completes the results will be buffered until `InstanceFlushIntervalMilliseconds` has elapsed or `InstanceWriteBufferSize` buffered writes have been made.
+
+The buffered writes are ordered by the time of the write using a single `insert ... on duplicate key update ...` call.  If the same host appears twice only the last write will be written to the database for that host.
+
+`InstanceFlushIntervalMilliseconds` *should be* well below `InstancePollSeconds` as making this value too high will mean the data is not being written to the orchestrator db backend.  This can lead to `not recently checked` problems.  Also the different health checks are run against the backend database state so not updating it frequently enough could lead to Orchestrator not detecting the different failure scenarios correctly.
+
+Suggested values to start with for larger Orchestrator environments might be:
+```
+  ...
+  "BufferInstanceWrites": true,
+  "InstanceWriteBufferSize": 1000, 
+  "InstanceFlushIntervalMilliseconds": 50,
+  "DiscoveryMaxConcurrency": 1000,
+  ...
+```
diff --git a/docs/configuration.md b/docs/configuration.md
index 083119ace3f22a89f0cd7a88876b0b564c67c423..2ee971706bec401c33bdce0b5fde049172f81de0 100644
--- a/docs/configuration.md
+++ b/docs/configuration.md
@@ -19,6 +19,7 @@ Use the following small steps to configure `orchestrator`:
 - [Raft](configuration-raft.md): configure a [orchestrator/raft](raft.md) cluster for high availability
 - Security: See [security](security.md) section.
 - [Key-Value stores](configuration-kv.md): configure and use key-value stores for master discovery.
+- [Hints on some settings suitable for larger orchestrator environments](configuration-large.md)
 
 ### Configuration sample file