QVD Header

Every QVD file starts with an XML header. It describes the table as a whole, each of its fields, and where the data of the table is located in the binary part of the file.

The root element QvdTableHeader contains general information about the table, one QvdFieldHeader per field and the position of the index table.

Example

An abridged header of a table with two fields and 1,000 records:

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<QvdTableHeader>
  <QvBuildNo>50689</QvBuildNo>
  <CreatorDoc>5b7e6c22-0d2a-44e7-8c55-c345c8760cfc</CreatorDoc>
  <CreateUtcTime>2026-01-05 14:37:21</CreateUtcTime>
  <SourceCreateUtcTime></SourceCreateUtcTime>
  <SourceFileUtcTime></SourceFileUtcTime>
  <SourceFileSize>-1</SourceFileSize>
  <StaleUtcTime></StaleUtcTime>
  <TableName>TestData</TableName>
  <Fields>
    <QvdFieldHeader>
      <FieldName>ID</FieldName>
      <BitOffset>0</BitOffset>
      <BitWidth>10</BitWidth>
      <Bias>0</Bias>
      <NumberFormat>
        <Type>INTEGER</Type>
        <nDec>0</nDec>
        <UseThou>1</UseThou>
        <Fmt>###0</Fmt>
        <Dec>,</Dec>
        <Thou></Thou>
      </NumberFormat>
      <NoOfSymbols>1000</NoOfSymbols>
      <Offset>0</Offset>
      <Length>5000</Length>
      <Comment></Comment>
      <Tags>
        <String>$numeric</String>
        <String>$integer</String>
      </Tags>
    </QvdFieldHeader>
    <QvdFieldHeader>
      <FieldName>Category</FieldName>
      <!-- ... -->
    </QvdFieldHeader>
  </Fields>
  <Compression></Compression>
  <RecordByteSize>2</RecordByteSize>
  <NoOfRecords>1000</NoOfRecords>
  <Offset>5027</Offset>
  <Length>2000</Length>
  <Lineage>
    <LineageInfo>
      <Discriminator>AUTOGENERATE;</Discriminator>
      <Statement></Statement>
    </LineageInfo>
  </Lineage>
  <Comment></Comment>
  <EncryptionInfo></EncryptionInfo>
  <TableTags></TableTags>
  <ProfilingData></ProfilingData>
</QvdTableHeader>

Encoding

The header is a well-formed XML document encoded in UTF-8. Numeric values are written as decimal text. Elements without a value are written as an empty element, e.g. <Comment></Comment>.

In all observed files, the elements appear in the order shown above.

Table header

The direct children of QvdTableHeader:

Element	Content	Description
`QvBuildNo`	integer	Build number of the software that wrote the file.
`CreatorDoc`	text	Identifier of the document that wrote the file. Observed as a GUID.
`CreateUtcTime`	timestamp	Time of creation in UTC, observed as `YYYY-MM-DD hh:mm:ss`.
`SourceCreateUtcTime`	text	Observed empty.
`SourceFileUtcTime`	text	Observed empty.
`SourceFileSize`	integer	Observed as `-1`.
`StaleUtcTime`	text	Observed empty.
`TableName`	text	Name of the table.
`Fields`	elements	One `QvdFieldHeader` per field, in field order.
`Compression`	text	Observed empty, see Compression.
`RecordByteSize`	integer	Size of one record of the index table in bytes.
`NoOfRecords`	integer	Number of records, i.e. rows of the table.
`Offset`	integer	Start of the index table, see Positions.
`Length`	integer	Size of the index table in bytes.
`Lineage`	elements	Origin of the data, see Lineage.
`Comment`	text	Comment on the table.
`EncryptionInfo`	–	Observed empty.
`TableTags`	–	Observed empty.
`ProfilingData`	–	Observed empty.

Field header

Each field of the table is described by a QvdFieldHeader inside Fields. The order of the QvdFieldHeader elements is the order of the columns of the table.

Element	Content	Description
`FieldName`	text	Name of the field.
`BitOffset`	integer	Position of the field’s index within a record, in bits.
`BitWidth`	integer	Number of bits of the field’s index within a record.
`Bias`	integer	Value added to the stored index, used to represent NULL.
`NumberFormat`	elements	Display format of the field’s values, see Number format.
`NoOfSymbols`	integer	Number of distinct values in the field’s symbol table.
`Offset`	integer	Start of the field’s symbol table, see Positions.
`Length`	integer	Size of the field’s symbol table in bytes.
`Comment`	text	Comment on the field.
`Tags`	elements	Tags of the field, see Tags.

How BitOffset, BitWidth and Bias are used is described on the Index table page; the content of the symbol table on the Symbol tables page.

Number format

NumberFormat describes how the values of a field are meant to be displayed. It does not affect how the values are stored: the values themselves are always taken from the symbol table.

Element	Content	Description
`Type`	text	Kind of format, see below.
`nDec`	integer	Number of decimal places.
`UseThou`	`0` or `1`	Whether a thousands separator is used.
`Fmt`	text	Format pattern, e.g. `###0`, `###0,00` or `DD.MM.YYYY`.
`Dec`	text	Decimal separator.
`Thou`	text	Thousands separator.

Observed values of Type are UNKNOWN, INTEGER, FIX and DATE. Software that writes QVD files also uses REAL, MONEY, TIME, TIMESTAMP, INTERVAL and ASCII.

Positions of the binary data

The header contains Offset and Length elements at two levels:

In each QvdFieldHeader, they give the position and size of that field’s symbol table.
Directly in QvdTableHeader, after Fields, they give the position and size of the index table.

All offsets are counted from the start of the binary part, i.e. from the byte following the NUL terminator of the header, not from the start of the file:

position in file = size of the header in bytes + 1 + Offset

In all observed files, the symbol tables of all fields are stored without gaps in field order, and the index table follows directly after the last symbol table. Its Length equals NoOfRecords × RecordByteSize.

Compression

The Compression element is empty in all observed files, and the binary part is stored uncompressed.

Lineage

Lineage describes where the data of the table came from. It contains any number of LineageInfo elements:

Element	Content	Description
`Discriminator`	text	Source of the data, e.g. `AUTOGENERATE;` for generated data.
`Statement`	text	Statement that loaded the data.

Licensed under CC BY 4.0.

Not affiliated with or endorsed by Qlik. Qlik and QVD are trademarks of their respective owners.