SSIS Excel Problem

We have the following situation:

Imagine that you are working for large finance organization and would like to load the year end balances into SQL Server database. The excel file look like this.

So we create a simple SSIS package using Excel connector and Ole DB destination, before loading the data we hit preview button.

For some reason last 3 lines have null values, This is because our connection string has not got IMEX=1

Provider=Microsoft.Jet.OLEDB.4.0;Data Source=C:\DEMO\Excel\mixed_data_types.xls;Extended Properties=“EXCEL 8.0;HDR=YES”;

We correct it

Provider=Microsoft.Jet.OLEDB.4.0;Data Source=C:\DEMO\Excel\mixed_data_types.xls;Extended Properties=“EXCEL 8.0;HDR=YES;IMEX=1”;

That looks much better so we load the data. Later the same day, the finance manager says that the balance is wrong you look at the data and there are some null values again.

It is a common problem: trying to load the data from Excel file half of the data is coming as nulls, or columns with more than 255 characters are truncated

Our source file has mixed datatypes some account codes have only numbers and some have characters as well. We have more cells with numbers only so Excel ODBC/Jet assumes the field type is numeric and shows the rest of it as nulls.

Setting IMEX=1 tells the driver to use Import mode. In this state, the registry setting ImportMixedTypes=Text will be noticed. This forces mixed data to be converted into text. For this to work reliably, you may also have to modify the registry setting, TypeGuessRows=8. The ISAM driver by default looks at the first eight rows and from that sampling determines the datatype. If this eight row sampling is all numeric, then setting IMEX=1 will not convert the default datatype to Text; it will remain numeric.

As partially explained here

ODBC/MS Jet scans first TypeGuessRows to determine field type

(TypeGuessRows=8 IMEX=1)

In your eight (8) scanned rows, if the column contains five (5) numeric values and three (3) text values, the provider returns five (5) numbers and three (3) null values.

In your eight (8) scanned rows, if the column contains three (3) numeric values and five (5) text values, the provider returns three (3) null values and five (5) text values.

In your eight (8) scanned rows, if the column contains four (4) numeric values and four (4) text values, the provider returns four (4) numbers and four (4) null values.

In your eight (8) scanned rows all of them less than 255 characters the provider will truncate all data to 255 characters

In your eight (8) scanned rows, if the column contains five (5) values with more length than 255 the provider will return more than 255 characters

We can change TypeGuessRows and set it to 0. This works very well for files up to 16384 lines.

The valid range of values for the TypeGuessRows key is 0 to 16. However, if the value is 0, the number of source rows scanned is 16384.
Read this Microsoft KB article

For larger files, it may not work again, so the best solution to the problem is not to use excel as data source but use text files instead. But this may bring another problem if user open CSV after it was exported it may loose leading zeros.

  1. Set IMEX=1 in the connection string
  2. Close any programs that are running.
  3. On the Start menu, click Run. Type Regedit and click OK.
  4. In the Registry Editor, expand the following key depending on the version of Excel that you are running:
    1. Excel 97
    2. Excel 2000 and later versions
  5. Select TypeGuessRows and on the Edit menu click Modify.
  6. In the Edit DWORD Value dialogue box, click Decimal under Base.
  7. Set the value to 1
  8. Open Excel file
  9. Make sure that the cells in the first line of the table have relevant data for example
    1. mixed numbers and text characters for text fields
    2. only numbers for numeric fields
    3. If some of the data will be longer than 255 characters make sure that first line cell has more 255 characters otherwise it will be truncated
This solution applies to all versions of MS Excel ODBC driver, Ole DB, MS Jet, .NET, DTS and SSIS

We have spent an enormous amount of time trying to get it fixed. So far we were not able to find a better solution.

The way Excel import works make it not possible to automate it. You have to modify most of excel files manually in order to load them.

This why we are no longer using ODBC/OleDB/Ms Jet for Excel connections. Our ETL solutions work correctly with Excel all the time. Try it yourself now.

  • Works directly with Excel 3.0-2007
  • No ODBC, OleDB or MS Jet Required
  • Works correctly with mixed data types
  • Works correctly with cells with more 255 characters
  • No need for IMEX=1, HDR=Yes or Registry hacks (TypeGuessRows)
  • Loads data correctly all the time + no need to edit Excel file
  • Can create Excel files in Excel 3.0-2007 format
  • Can insert data starting from specific cell
  • Can clear area before adding data into Excel
  • Can add headers
  • knowledgebase/ssis_excel_problem.txt
  • Last modified: 13/04/2019 17:24
  • by admin