In terms of improving the group import process in NetexCloud, I wish to flag on what happened in the pre-production environment after launching an import of 64 groups:
The summary perceived by the end user is as follows:
slow process: total time in accomplishment of all the work 55 minutes, for 64 registers... What will happen with 100?
Unreliability in what the screen says: the import was done in the cloud instantly and 62 groups were created in the LMS Pack in the first minute. The remaining 2 groups took 54 minutes to propagate.
After reviewing the process with the parties, I summarize the operation that currently follows:
- For each cloud group create the group in database and create a message in the queue to send the info to central
- A processor of the queue takes the message and generates request to central (I don't know if it makes checks before launching the creation operation)
- Central collects the request and generates the group in database.
Problem that arose in this import:
In central deadlocks are produced due to possible multiple requests for several threads from the cloud.
Proposal for improvement:
- Instead of generating X messages in a queue, it would be more logical to send all the information collected in a single message. If this fails, it will be reattempted later.
- Each platform manages the information at its own convenience, which is why it requires it to implement a web service so that it can receive the information it needs.
- Communications between platforms the fewer requests there are between them the faster the communication between them.
Customer support service by UserEcho