On MS SQL Server, during the night we like to update the statistics, but this process is taking a long time and runs well into the morning.
Is it okay sample by rows where the number of rows is +100,000? or will this give the wrong statistics?
On MS SQL Server, during the night we like to update the statistics, but this process is taking a long time and runs well into the morning. Is it okay sample by rows where the number of rows is +100,000? or will this give the wrong statistics? 


Are you current using a FULLSCAN or PERCENT? How are you queries performing under your current scheme? It isn't necessarily easy but from my experience tying the two, UPDATE STATISTICS to query performance, provides a more arguable position for changing. Presumably the running well into the morning is causing issues. Whether or not a row count is "Okay" is a bit like the car ad ("your mileage may vary"). 


Your question cannot be answered by sheer volume of rows. You need to consider selectivity and density for a start. Take a look at this article for more information: 

