Limit on SQLDataReader w/ Blobs

[Garth Wells]

I use a filtered SELECT to populate the SQLDataReader (rdr) with
a filename and a blob (pdf). I then use File.WriteAllBy tes to write
each pdf to disk.

----------------------------------------
rdr = command.Execute Reader();

while (rdr.Read())
{
byte[] BinaryImage = (byte[])rdr["attachment_fil e"];
File.WriteAllBy tes("\\" + rdr["fn"].ToString(), BinaryImage);
}
----------------------------------------

I tested the code on a small resultset and it works as expected. My
client tried it on the live database (I can't get acccess to this database),
and he said he thinks it has some built-in upper limit on the number of
rows it can process because he doesn't think it writes all files that are
returned by the SELECT. Instead of adding rows to my test database
I was hoping someone might be able to provide insight on any upper
limit this approach might have and suggest a workaround.

Thanks

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

Garth Wells wrote:

I use a filtered SELECT to populate the SQLDataReader (rdr) with
a filename and a blob (pdf). I then use File.WriteAllBy tes to write
each pdf to disk.

----------------------------------------
rdr = command.Execute Reader();

while (rdr.Read())
{
byte[] BinaryImage = (byte[])rdr["attachment_fil e"];
File.WriteAllBy tes("\\" + rdr["fn"].ToString(), BinaryImage);
}
----------------------------------------

I tested the code on a small resultset and it works as expected. My
client tried it on the live database (I can't get acccess to this database),
and he said he thinks it has some built-in upper limit on the number of
rows it can process because he doesn't think it writes all files that are
returned by the SELECT. Instead of adding rows to my test database
I was hoping someone might be able to provide insight on any upper
limit this approach might have and suggest a workaround.

An SQLServer BLOB (called IMAGE) can be up to 2 GB.

There is lots of things that can prevent you from reading
such a big data item in a single call.

You can try using a loop and via rdr.GetBytes read 100 KB or 1 MB
a time and write them to the file.

Arne

[Marc Gravell]

You can try using a loop and via rdr.GetBytes read 100 KB or 1 MB

a time and write them to the file.

Additional: when handling large BLOBs, it would be a good idea to set
CommandBehavior .SequentialAcce ss in the reader; this will allow it to
work more efficiently, as it doesn't try to load the entire row into
memory - just the bytes you are reading at that moment. Of course,
with the GetBytes() approach you are forced to use the slightly longer
version of writing, using a FileStream.

called IMAGE

Or varbinary(max) ;-p

Marc

[=?ISO-8859-1?Q?Arne_Vajh=F8j?=]

Marc Gravell wrote:

>called IMAGE

Or varbinary(max) ;-p

Not by people of my age.

:-)

Arne

[Garth Wells]

Garth Wells wrote:

>I use a filtered SELECT to populate the SQLDataReader (rdr) with
a filename and a blob (pdf). I then use File.WriteAllBy tes to write
each pdf to disk.

----------------------------------------
rdr = command.Execute Reader();

while (rdr.Read())
{
byte[] BinaryImage = (byte[])rdr["attachment_fil e"];
File.WriteAllBy tes("\\" + rdr["fn"].ToString(), BinaryImage);
}
----------------------------------------

I tested the code on a small resultset and it works as expected. My
client tried it on the live database (I can't get acccess to this database),
and he said he thinks it has some built-in upper limit on the number of
rows it can process because he doesn't think it writes all files that are
returned by the SELECT. Instead of adding rows to my test database
I was hoping someone might be able to provide insight on any upper
limit this approach might have and suggest a workaround.

An SQLServer BLOB (called IMAGE) can be up to 2 GB.

There is lots of things that can prevent you from reading
such a big data item in a single call.

You can try using a loop and via rdr.GetBytes read 100 KB or 1 MB
a time and write them to the file.

I reviewed all the file sizes and the largest was 4 MB. There must be
a max SQLDataReader size, but a quick google did not indicate the
value.

[Marc Gravell]

It is conceivable that the limit is in the line:
rdr["attachment_fil e"];

For BLOBs, I would be using GetBytes() successively with increasing
offsets (into a re-usable buffer)

Marc

[Garth Wells]

It is conceivable that the limit is in the line:

rdr["attachment_fil e"];

For BLOBs, I would be using GetBytes() successively with increasing
offsets (into a re-usable buffer)

Can you point me to an example of this approach?

Thanks

[Marc Gravell]

OK: http://weblogs.asp.net/cazzu/archive.../27/25568.aspx

The context here is streaming an image from the database to a browser
- however ,the BLOB reading is the same.

Personally, I think the "read" step could be neater - it makes the not-
strictly-true assumption that a buffer must be filled; I'd go with:
while ((size = r.GetBytes(1, idx, buffer, 0, ChunkSize)) 0)
{
context.Respons e.BinaryWrite(b uffer, 0, size);
idx += size;
}

(and remove the "last bytes" stuff down to the end-brace, since we
have dealt with them above).

"Replace context.Respons e" with your own stream / writer...

Marc