Using 8.0.8. Our Store and Forward disk cache is extremely slow. If we disable the disk cache, we can enter records from our tags into our MSSQL 2016 enterprise database at a minimum rate of 5000 transactions/sec. Transactions / sec are usually in the 1000-2000 range. When we enable our ignition local disk cache, the store and forward disk cache transactions go quickly as well, but only as long as the disk cache is virtually empty. Turning on the ‘force use of disk cache’ option will eventually cause too many records to build up in the cache, and our forwarding rate will slow to 0-400 transactions / sec. This causes more records to build up, and eventually the rate of insertion slows to zero.
The work-around is to disable the store and forward disk cache, and we have a new database connection set up with one of these old caches, slowly draining into the database at a rate of about 30 transactions / sec. This causes deadlock quarantines, but those can be retried, so at least we can get our data eventually.
I have a ticket in with support, but it is not moving and we’re stuck with a problem where if our sql server is rebooted, we’re going to end up losing transactions. The memory buffer is only big enough to store about 20 seconds of data, and support recommended that we decrease the size of this buffer (not sure why). This isn’t long enough for our SQL server to reboot fully, especially if the server requires updates (Windows).
Anybody run into this issue, where your server and database can handle the load but the local store & forward disk cache isn’t able to forward the transactions fast enough to keep up?