Hi, we have published a number of map services and they all point to feature classes in a Microsoft SQL Server database. My question is, if the map services are not being consumed by anything (e.g. apps, ArcMap, Portal... etc), would the map services still put any load/work on the database (not ArcGIS Server)? For instance, would the map services continuously to ping the database? The only thing I can see is there is a "Periodically check and repair data connection for idle instances" setting that will ping the database periodically (this setting can be turned off).
The map services are all quite simple/standard in that they simply point to the feature classes, some might have query definitions, that's about it (no extra stuff like WMS, feature access capabilities etc).
Same goes for the feature classes, they are straight-forward feature classes, nothing fancy like network datasets, domains, versionings... etc.
ArcGIS is 10.5 running on Windows 2012 R2; MS SQL is version 2014 SP2.
As I understand it; if any instances of the map service are running, i.e. minimum instances = 1, then yes there is a "live" connection. If no one is consuming the instance then it is more or less idle but is available for use.
Are there some issues with SQL Server side?
No issues with the SQL Server side. We are trying to setup a disaster-recovery / standby environment.
So the production database is called PRD-GIS-DB, the disaster-recovery database is called DR-GIS-DB (they automatically sync almost real-time). They are referenced by a common name called GIS-DB (we can't access DR-GIS-DB directly). During normal operations, GIS-DB points to PRD-GIS-DB. When disaster happens, GIS-DB gets manually switched to DR-GIS-DB.
Now when I publish map services on the disaster-recovery environment (which are blocked to be accessed during normal operations), the map services' SDE connections will need to use GIS-DB (because we can't use DR-GIS-DB directly). During normal operations, the disaster-recovery map services would be pointing to PRD-GIS-DB. So I am wondering if I should switch off these disaster-recovery map services if they put on any load on PRD-GIS-DB (and switch them on manually when disaster happens), or simply leave them there (clients are blocked from accessing these during normal operations) so when disaster happens, the DBAs just need to switch GIS-DB to DR-GIS-DB and no need to turn on the map services.
Have you reviewed this doc: Connections to highly available SQL Server databases—Help | ArcGIS Desktop
You could put the number of instances on the DR map services to "0" till they are needed. If the DBA's have the SQL Server configured the switch should happen normally without intervention.
Thanks for the link, it helped. For anyone else viewing this thread, our GIS enterprise setup is this one: Single-machine high-availability (active-passive) deployment—ArcGIS Server (Windows) Installation Gu...