Looking for BI performance standards (response time)
4 posters
Page 1 of 1
Looking for BI performance standards (response time)
We are trying to establish bench mark for BI report performance. Is there any Industry Standards on BI reports response time, like
• What is the average response time for an average standard report.
• Average response time for drill down/drill through query to shown up.
• Dashboard/Scorecard refresh time.
Any reference would be helpful. Thanks in advance.
• What is the average response time for an average standard report.
• Average response time for drill down/drill through query to shown up.
• Dashboard/Scorecard refresh time.
Any reference would be helpful. Thanks in advance.
Rajesh- Posts : 1
Join date : 2011-01-18
Re: Looking for BI performance standards (response time)
Other than it being better than the old system, there isn't a standard or benchmark that I am aware of. There are far too many variables to have any number mean anything.
5 seconds
My target is always 5 seconds - "Analysis At the Speed of Thought".
If any of those actions take longer than that, the user often gets frustrated and either gives up discouraged or cancels and retries. When a user is analysing, longer delays mean they will lose the mental map of what they are investigating.
That said, it's a "target", so I don't get too fussed about the performance of extreme cases (e.g. users using Excel PivotTables on SSAS cubes as a data dump tool for thousands of rows, or users with weak PCs / network connections).
If any of those actions take longer than that, the user often gets frustrated and either gives up discouraged or cancels and retries. When a user is analysing, longer delays mean they will lose the mental map of what they are investigating.
That said, it's a "target", so I don't get too fussed about the performance of extreme cases (e.g. users using Excel PivotTables on SSAS cubes as a data dump tool for thousands of rows, or users with weak PCs / network connections).
Re: Looking for BI performance standards (response time)
Be realistic and have eyes on following parameters :
1) Time of the day report is being run.
2) From which locations reports will run.
3) Whether a report is data hungry.
4) Bandwidth of the internet/intranet available.
------------
If using cubes then design some reports which are readymade and can directly give the data.
If few reports needed every now and then , schedule those reports so that in morning when users come they get the direct report output.
More in depth analysis is propotional to more number of queries and data requirments.
Set the expectations right in the beginning else .... difficult in the end.
1) Time of the day report is being run.
2) From which locations reports will run.
3) Whether a report is data hungry.
4) Bandwidth of the internet/intranet available.
------------
If using cubes then design some reports which are readymade and can directly give the data.
If few reports needed every now and then , schedule those reports so that in morning when users come they get the direct report output.
More in depth analysis is propotional to more number of queries and data requirments.
Set the expectations right in the beginning else .... difficult in the end.
Vishy- Posts : 75
Join date : 2012-02-28
Location : India
Similar topics
» Adding new fields to a live system
» Store the tax rate as the attribute of product in product dimension table?
» Combined Date Time dimension vs. Separate Day and Time dimensions and timezones
» Time dimension design for cross time zone and custom calendar warehouse, 1min granularity
» Does a SCD Type 1 Change Response Always Update All Historical Records?
» Store the tax rate as the attribute of product in product dimension table?
» Combined Date Time dimension vs. Separate Day and Time dimensions and timezones
» Time dimension design for cross time zone and custom calendar warehouse, 1min granularity
» Does a SCD Type 1 Change Response Always Update All Historical Records?
Page 1 of 1
Permissions in this forum:
You cannot reply to topics in this forum