24-05-23 12:47 PM
Hi,
I have some experience via manual, automatic and AutomateC archiving and deletion. I always opt for AutomateC since from an efficiency perspective I do not have to tie up a RR for the action.
I am sure I need to know the specs of the servers involved for the following questions but assuming I had that:
Assuming I had 2 years of data and I wanted to cut it down to 180 days, Is the server load more on a longer deletion frame or is it about the same as a shorter frame (just extended time due to a larger payload)? I just want to know here if there is even a need to request downtime or if I can just choose a low volume time.
I recently did a development environment test with sub 100GB of data deleted (client prefers deletion to archiving in this scenario) and it took longer than I anticipated. But going into production I want to provide a maintenance window to our client. How does one go about estimating a time? I imagine it will be a deletion of about 350GB.
24-05-23 03:25 PM
Hi Andrew,
Your business requirements will dictate the requirements for data retention, but from a best-practice standpoint we typically recommend our customers to always set aside time for regular downtime and maintenance windows to ensure Blue Prism environments are kept in optimal condition. Our general guidelines for database maintenance can be found in the documentation linked within this KB article here.
It's important to note that Blue Prism databases are transactional in nature and are not intended to be used as a repository for historical data as doing so can lead to performance issues, SQL timeouts, transaction deadlocks, etc. This is why our "Housekeeping Scripts" by default only keep data that is 7 days old or newer, with the recommendation to use Archiving or Data Gateways functionality to regularly move off historical data needed for reporting purposes to another storage location (rather than the live BP database performing transactions).
To address your question about estimating a downtime window, one strategy could be to work with your DBA to take a backup of your current Prod database and copy it into one of your lower environments (Dev/Test/etc) that are similar to your Prod infrastructure, and perform the required maintenance there to determine the duration. You could then use this information to better determine an appropriate downtime window for Prod DB maintenance and re-booting of BPE components.