Performance comparison between JXL and POI FOR excel file generations

Question

Here is the code that generates .xls file using JXL:

public void generateXls(String fileName, int sheets, int cols, int rows) {

    if (cols > 256) {
        throw new IllegalArgumentException("Error: number of cols per sheet must be < 256");
    }

    if (rows > 65536) {
        throw new IllegalArgumentException("Error: number of rows per sheet  must be < 65536");
    }

    String fullName = fileName + ".xls";
    WritableWorkbook workbook = null;
    try {
        workbook = Workbook.createWorkbook(new File(fullName));
    } catch (IOException e) {
        e.printStackTrace();
    }
    Random random = new Random();
    for (int s = 0; s < sheets; s++ ) {
        WritableSheet sheet = workbook.createSheet("Sheet" + s, 0);
        for (int i = 0; i < cols ; i++) {
            for (int j = 0;  j < rows; j++) {
                Number number = new Number(i, j, random.nextDouble()*1000);
                try {
                    sheet.addCell(number);
                } catch (RowsExceededException e) {
                    throw new RuntimeException("Error: too many rows in a sheet");
                } catch (WriteException e) {
                    throw new RuntimeException("Error occured while adding cell to sheet", e);
                }
            }
        }
    }

    try {
        workbook.write(); 
        workbook.close();
    } catch (WriteException e) {
        e.printStackTrace();
    } catch (IOException e) {
        e.printStackTrace();
    }

}

Here is the code that generates .xls and .xlsx files using POI:

public void generateXlsx(String fileName, int sheets, int cols, int rows) {

    if (cols > 16383) {
        throw new IllegalArgumentException("Error: number of cols per sheet must be < 16383");
    }
    XSSFWorkbook workbook = new XSSFWorkbook();

    Random random = new Random();
    for (int s = 0; s < sheets; s++ ) {
        XSSFSheet sheet = workbook.createSheet();
        for (int i = 0; i < rows ; i++) {
            XSSFRow row =  sheet.createRow(i);
            for (int j = 0;  j < cols; j++) {
                XSSFCell cell = row.createCell(j);
                cell.setCellValue(random.nextDouble()*1000);
            }
        }
    }

    FileOutputStream fileOut = null;
    try {
        fileOut = new FileOutputStream(fileName);
    } catch (FileNotFoundException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }
    try {
        workbook.write(fileOut);
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }

    try {
        fileOut.flush();
        fileOut.close();
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }


}

public void generateXls(String fileName, int sheets, int cols, int rows) {

    if (cols > 256) {
        throw new IllegalArgumentException("Error: number of cols per sheet must be < 256");
    }

    HSSFWorkbook workbook = new HSSFWorkbook();

    Random random = new Random();
    for (int s = 0; s < sheets; s++ ) {
        HSSFSheet sheet = workbook.createSheet();
        for (int i = 0; i < rows ; i++) {
            HSSFRow row =  sheet.createRow(i);
            for (int j = 0;  j < cols; j++) {
                HSSFCell cell = row.createCell(j);
                cell.setCellValue(random.nextDouble()*1000);
            }
        }
    }

    FileOutputStream fileOut = null;
    try {
        fileOut = new FileOutputStream(fileName);
    } catch (FileNotFoundException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }
    try {
        workbook.write(fileOut);
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }

    try {
        fileOut.flush();
        fileOut.close();
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }


}

I wrote a performance test:

public static void main(String[] args) {

    int sheets =1;
    int cols = 255;
    int rows = 20000;
    long a1 = System.currentTimeMillis();
    ExcelFileGenerator generator  = new ExcelFileGenerator();
    generator.generateXls("xlsJXL.xls", sheets, cols, rows);
    long xls = System.currentTimeMillis()-a1;
    System.out.println("xlsJXL: " + xls);
    ExcelFileGeneratorPOI generatorPOI = new ExcelFileGeneratorPOI();
    long a2 = System.currentTimeMillis();
    generatorPOI.generateXls("xlsPOI.xls", sheets, cols, rows);
    long xlsPoi = System.currentTimeMillis()-a2;
    System.out.println("xlsPOI: " + xlsPoi);
    long a3 = System.currentTimeMillis();
    generatorPOI.generateXlsx("xlsxPOI.xlsx", sheets, cols, rows);
    long xlsx = System.currentTimeMillis()-a3;
    System.out.println("xlsxPOI: " + xlsx);
}

The results are: sheets =1 cols = 255 rows = 10 xlsJXL: 133 xlsPOI: 162 xlsxPOI: 645

sheets =1 cols = 10 rows = 255 xlsJXL: 130 xlsPOI: 140 xlsxPOI: 650

sheets =10 cols = 255 rows = 255 xlsJXL: 611 xlsPOI: 784 xlsxPOI: 16228

sheets =2 cols = 100 rows = 10000

xlsJXL: 2755 xlsPOI: 3270 xlsxPOI: Exception in thread "main" java.lang.OutOfMemoryError: Java heap space

any reason by creating .xlsx with POI is much slower than creating .xls?

Answer 1

xls is a binary based format, xlsx is an XML based format and requires more work to read/write.

xlsx might also require an in memory document model to parse/build the XML which could be more complex.

Finally, xls could be optimised better because it has been supported longer.