The old and new versions of SQL Server CE differences

Comparison of the difference between CE7 and 2014 CE12:

Table 1. Estimation column connections connected mode

Old columns connected CE for all those involved in the step-by-statistics estimate. The new estimate for CE only collect the maximum and minimum step statistics, while the value of non-uniform distribution of the connecting columns, old CE has a more accurate estimate.

2. For where multiple conditional statements estimates

CE that old computing respective conditions independent, there is no correlation, estimates the condition of the highest selectivity for the standard alone, the new CE condition estimation mode is a selective * sqrt (2 selective conditions) * sqrt (sqrt (3 selection conditions sexual)), a condition in which selective> condition 2> condition 3, this correlation estimates for multiple conditional statements, the new CE more accurate estimate.

3.where conditions beyond the boundaries of statistical information to estimate

CE beyond the old boundaries of statistical information to estimate where the conditions for the use of simple statistics boundary step value corresponding to an estimate, the total number of lines used by the new CE * way line density estimates more accurate.

4. In connection condition on an estimate of the many conditions equijoins

Old CE selective manner using a single simple condition column multiplied values ​​are often too small, a new use of a smaller column CE after two conditions are repeated multiply connected to the average density of the two tables, estimates larger .

The non-multi-condition estimation equivalent connection.

Old CE selectivity for simple connection conditions set by multiplying the estimated value is low, the new connection is assumed that CE is a small table-many relationship between the large table, an estimate of the input and large table is calculated, the estimated value is high.

6. The non-connected conditions are estimated:

Old CE conditions are that the non-connected independent of each other, a large selection condition and estimates that the column is connected at both ends at the same time meet the estimated value is high, the non-connection condition of the new CE different tables that are not related to selectively connect the first estimates, after application of the non-connected condition and then filtration column.

7. After connecting to repeat the operation (distinct, group by)

Old connected CE-many estimates before the de-emphasis, and then connect the smaller estimate the result after valuation, the new CE is first de-emphasis.

Guess you like

Origin www.cnblogs.com/database/p/11956479.html