I have a hosted map image layer in a spatiotemporal datastore with approx 700 mil rows. This service drew fine when there were less rows, but now it's getting chunky it is timing out when drawing in Portal, and is taking > 10 minutes to draw in Pro. The associated feature layer is drawing fine (sub 30 seconds) in both Pro and Portal.
This data was added to the datastore via GeoEvent server.
Does anyone have recommendations on spatiotemporal datastore configurations to better support very large map image layers?
Current config;
- 3 node cluster
- 16 cores + 64 GB RAM for one node
- 8 core + 64 GB RAM for other two nodes
- 5 shards
- replication factor = 1
- refresh interval = 60 seconds
- rolling data = monthly
- shrink = yes
- aggregation = hex's to level 12