The short answer - it depends on a variety of factors.
The complete answer is that One Model is an advanced analytical platform combining data from multiple systems and bringing together historical trends and events across potentially very large datasets.
Running a query in One Model is much more complicated than a simple transaction, such as looking up an employee in your HRIS, that you might perform in your regular business application.
The complexity of analytical queries grows based on the following factors:
- The number, and complexity, of metrics in the query.
- Whether there is a time function as part of the metric calculation (e.g. rolling periods).
- The number of time periods in the time dimension.
- The number of dimensions.
- The size of the underlying dataset.
- The complexity of the business logic in the underlying data model.
- Whether the underlying data model has the appropriate keys for the relationships.
- The complexity of the role-based permissions being applied to individual users.
- The number of tiles/ widgets on my storyboard.
Due to these complexities, One Model is built on a hyperscale platform that continually works on performance optimization, and aims to facilitate a super-fast end-user experience through the use of caching.
Caching works by shortening the time it takes to fetch your data. Every time a query is run in One Model, it is added to the cache, so if you or someone else runs the same query again it will be fetched from the cache and doesn’t need to go back to query the database again.
However, the cache is cleared each time your data is reprocessed (usually nightly), so we have also added an automated cache warming process that captures all of the Storyboard queries users are running (over a rolling 30 day period) and loads these into the cache after data processing is complete. [NB: This automated cache warming process only applies to Storyboards and doesn’t apply to ad-hoc queries run in Explore.]
Naturally for large organizations, this cache warming process can take a long time, so we are always looking for ways to improve this algorithm and platform performance.
How can I investigate a “slow” running query or Storyboard?
- Check your internet connection, firewall and VPN settings. For example, if you are outside of the United States (US) and are having difficulty accessing the US-based server or it is slow, but team members within the US are not having issues, then the issue may be with how you’re accessing the site. If the site is faster when you are accessing in the office versus from home, you may have a slow internet connection. Understanding your experience accessing the Storyboard versus colleagues can be helpful for troubleshooting.
- Consider the complexity of your query and the number of tiles on a Storyboard page. The best performing Storyboards typically hold six to eight tiles per page, and if you have to scroll down the Storyboard to see more tiles then consider spreading the tiles across multiple pages. (Please note that Storyboard Pages are coming soon).
- If you are an Admin user, use the Proxy feature to check the query/Storyboard performance other users will experience – this will also help assess the impact of role based permissions.
- Just make sure that the testing you are conducting is fully inclusive of all roles on the user account because if you are only testing a subset of roles, it will be a different experience.
- And, if you are using automatic assignment of roles, remember that new users will not have roles applied to them until they login.
- Consider whether filters should be applied to focus the query results. There is a filter prompt feature to Storyboards to help in cases where you want the end-user to choose some specific filters before opening a Storyboard.
- Digging deeper, take a copy of the query and rebuild it one piece at a time and see how the time grows for each element you are adding.
- You can take the same approach with a Storyboard and see at what point you are slowing things down.
- If you are still having trouble with a slow running query or Storyboard, please contact Customer Success for assistance.
Comments
0 comments
Please sign in to leave a comment.