In order to ensure adequate performance (reasonable load times) we suggest limiting scope in analysis configuration to ~100 million cells (cells = indices x tags = rows x columns). If you try larger analyses, pages will take longer to load and may eventually crash. You will not get an error that your analysis was too big, but the first thing you should check is the point scope of your analysis.
Let’s look at how this could play out for some typical base sizes:
| # Tags | Base Frequency | Period Encouraged |
|---|---|---|
| 4000 | 1 point/10 min | <6 months |
| 4000 | 1 point/1 min | <1 month |
| 1000 | 1 point/1 min | <2 month |
If you had >100 million cells in your initial analysis design, consider these approaches:
- Can you do the analysis in a lower frequency database (ie digital twin)? (Number of indices has a stronger effect on performance than the number of variables).
- Can you limit the input variables to sensically related tags using data properties or a data group?
- Can you choose a smaller period for analysis then use the Rule Report to assess findings on other periods of data?
Also note if others are performing analyses at the same time such as in a training session, performance for all is expected to decline. It is best to run only 1 analysis at a time in a team setting.