microsoft.public.sqlserver.programming Google Group

Re: Why can’t recursive queries contain...?

m...@nospam.com (Michael C) — Mon, 11 Nov 2009 03:01:04 UT

news:bda19704-b1ab-45fe-8e94-
SQL server is full of limitations that don't make sense. We've only recently
got the ability to do something as basic as define a variable of infinite
length string. When we got CLR integration we were given a limit of 8K of
memory to work with (E I G H T K!!!!!!). When we got the ability to define

Re: Sql Server on VMWare

kgbo...@earthlink.net (TheSQLGuru) — Mon, 11 Nov 2009 01:05:55 UT

If you need any performance at all I HIGHLY recommend you get an experienced
consultant in to help provision/configure your system.

Re: Transaction load

sqlmvpnooos...@shadhawk.com (Andrew J. Kelly) — Mon, 11 Nov 2009 00:35:48 UT

Mike,
The transaction log entries are crucial to peak performance when it comes to
writes. Essentially if one insert is held up due to bottlenecks on writing
to the log file so will all other transactions that will also be writing at
that time. The more concurrent writes you have the more the potential for a

Re: SQL 2008 - FILESTREAM OpenSQLFilestream invalid handle

nie_spamuj...@radio.kielce.pl (jh) — Sun, 11 Nov 2009 21:15:24 UT

Uzytkownik "Erland Sommarskog" <esq...@sommarskog.se> napisal w wiadomosci
OK. Here is declaration of OpenSqlFilestream:
<code>
function OpenSqlFilestream(
Path: PChar;
Access: Cardinal;
Options: Cardinal;
txnToken: TtxnToken;
txnTokenLenght: Cardinal;
AllocationSize: PInt64): THandle stdcall; external 'sqlncli10.dll';

Re: creating table on secondery filegroup

please_reply_on_newsgro...@domain.com (Kalen Delaney) — Sun, 11 Nov 2009 21:00:28 UT

In SQL 2008, you can INSERT into an existing table, using a nested SELECT,
with the same performance as SELECT INTO (in many cases). So you can create
the table on your preferred filegroup, and then copy the rows from the other
table into it. Check the BOL for full details.

Re: SQL 2008 - FILESTREAM OpenSQLFilestream invalid handle

esq...@sommarskog.se (Erland Sommarskog) — Sun, 11 Nov 2009 20:07:31 UT

I suspect this is a "how do I do this in Delphi", which is not easy to
assist with, since I don't know Delphi. I would guess that you have some
error around one of the parameters to OpenSqlFilestream or the
import of it.
Could you post the code, including how you import OpenSqlFilestream?
No promises, but maybe we are able to spot something.

Re: Transaction load

esq...@sommarskog.se (Erland Sommarskog) — Sun, 11 Nov 2009 19:09:17 UT

More or less. BULK INSERT uses OLE DB, and below the covers there are
is a command INSERT BULK, which you can see in a Profiler trace.
But keep in mind that while they are similar, they are to some extent
different implementations, and there are subtle differences. Sometimes
BCP accepts a file that BULK INSERT barfs at, or vice versa.

Re: creating table on secondery filegroup

guzma...@nospam-online.sbcglobal.net (Dan Guzman) — Sun, 11 Nov 2009 18:22:08 UT

SELECT...INTO creates the table on the default filegroup. You'll need to
mark the secondary filegroup as the default one beforehand:
ALTER DATABASE MyDatabase
MODIFY FILEGROUP [MySecondaryGroup] DEFAULT;

Re: Sql Server on VMWare

remove_before_replying_dpor...@acm.org (David Portas) — Sun, 11 Nov 2009 18:15:27 UT

[link]
[link]

Re: Transaction load

_no_s...@_no_spam.com (Mike23) — Sun, 11 Nov 2009 17:06:46 UT

Thanks. I am aware of this but was just speculating that "SqlBulkCopy" is
probably a wrapper for BULK INSERT or some other high-volume (native) SQL
Server statement (possibly internal to MSFT only). My original issue however
is whether running "SqlBulkCopy" (or BULK INSERT directly) will have an

creating table on secondery filegroup

r...@top.com (Roy Goldhammer) — Sun, 11 Nov 2009 16:54:32 UT

hello there
i have database with two filegroups and for each filegroup there is diffrent
file name
How can I use Select .. into and create new table on secondery filegroup?

Re: Transaction load

guzma...@nospam-online.sbcglobal.net (Dan Guzman) — Sun, 11 Nov 2009 15:12:29 UT

BULK INSERT is a Transact-SQL statement whereas the SqlBulkCopy class (and
its unmanaged ODBC/OLEDB cousins) are native client APIs. Unlike BULK
INSERT, the client APIs allow you to load directly from program variables.

Re: Transaction load

_no_s...@_no_spam.com (Mike23) — Sun, 11 Nov 2009 12:48:17 UT

Thanks. Can you also briefly elaborate on my last response to him (just
posted). Appreciate your help.
Not familir with this but probably not necessary for my needs unless it
solves some problem specificically related to this issue (resource
exhaustion). The bulk insert I tested ("SqlBulkCopy" in .NET) is very fast

Re: Transaction load

_no_s...@_no_spam.com (Mike23) — Sun, 11 Nov 2009 12:34:21 UT

Thanks for you help and I agree. It's hard to appreciate when you're not
really familiar with the details. In this case however it's a fairly simple
situation. Assuming 10,000 customers in the short-term, 100,000 long term,
and each customer will be uploading a single file just once every few weeks

Re: Transaction load

_no_s...@_no_spam.com (Mike23) — Sun, 11 Nov 2009 12:09:16 UT

Thanks. I'm actually using the "SqlBulkCopy" class (assuming you're familiar
with .NET) and this presumably wraps BULK INSERT (still researching things).
I also took a quick look at ROWS_PER_BATCH and this may prove useful (thanks
for pointing this out). If possible however, I'd like to insert all data in