I am trying to understand the variances I am seeing in response times for Laserfiche JRA API calls to a repository. I am not just talking about searches which I can understand will vary based on the complexity of the query and the amount of data returned but I am also seeing variances in time just obtaining a session. This seems to vary anywhere from 36ms to 5s. We have used Wireshark to try to determine where the time is being spent and could see that most of the time waiting for Laserfiche server to send back a response.
Through our testing we realised that search queries don't use the search engine. Is this because it goes directly to the SQL server instead? Can someone explain to me what resources/services the Laserfiche API uses to help me identity what could be affecting the response time? Is there other things that run on the Laserfiche server eg. import agent that could affect the time responses? When I run the same query in the client it is lightning fast compared to what I am seeing via the API. Can someone explain to me why there is such a big difference other than time transferring data back to the server requesting the information?
The performance of search queries has deteriorated over time despite the queries being run not having changed. I am not sure if this is a consequence of upgrading the Laserfiche version or whether something else at play. Any clues to this would be greatly appreciated.