In a two node, SQL AlwaysOn cluster with synchronous replication, we have seen the following problems periodically:
• Initial sync of a new added database with automatic seeding is periodically very slow.
• Database replication is periodically very slow.
In both scenarios, the fix is to restart the AlwaysOn replication endpoint.
We use this command: “alter endpoint Hadr_endpoint state = stopped” and “alter endpoint Hadr_endpoint state = started”
We have tested the speed reduction with “HammerDB” software.
When the synchronization is good, we see around 1 million TPM, when it´s slow it´s about 25.000 TPMs.
Speed is going up if we change the synchronization mode to asynchronous commit and goes back down when we change it back to synchronous commit.
The Cluster is SQL version 2017 with CU17 update on Windows server 2016 operating system. It´s running on dedicated VmWare servers. Each SQL server have 12 CPUs and 1.5TB Ram. The CPU is an Intel Xeon E5-2643 V4 @ 3,40GHz. There is a dedicated and isolated network for synchronization. Storage is Dell EMC XtremIO all flash array.
we don´t know what the cause is or how to trigger or fix it.
Could you please help
• Initial sync of a new added database with automatic seeding is periodically very slow.
• Database replication is periodically very slow.
In both scenarios, the fix is to restart the AlwaysOn replication endpoint.
We use this command: “alter endpoint Hadr_endpoint state = stopped” and “alter endpoint Hadr_endpoint state = started”
We have tested the speed reduction with “HammerDB” software.
When the synchronization is good, we see around 1 million TPM, when it´s slow it´s about 25.000 TPMs.
Speed is going up if we change the synchronization mode to asynchronous commit and goes back down when we change it back to synchronous commit.
The Cluster is SQL version 2017 with CU17 update on Windows server 2016 operating system. It´s running on dedicated VmWare servers. Each SQL server have 12 CPUs and 1.5TB Ram. The CPU is an Intel Xeon E5-2643 V4 @ 3,40GHz. There is a dedicated and isolated network for synchronization. Storage is Dell EMC XtremIO all flash array.
we don´t know what the cause is or how to trigger or fix it.
Could you please help
Kiran