|
The current TPC-C[^] raw performance benchmarks suggest that Oracle on HP/UX on outlandish hardware (64-way Itanium2 1.5GHz, 512Gb of memory) does currently outperform SQL Server 2000 (64-bit) on the same outlandish hardware (look at the second and third rows, the first had 1024Gb of RAM).
However, most of us are lucky to be using 512 megabytes rather than 512 gigabytes.
Frankly, SQL Server usually performs well enough on almost any hardware. Oracle really requires massive amounts of memory - 9i couldn't create a database on our test system with 192Mb (taking 6 hours and then the client crashed), while SQL Server 2000 creates a new database in about 5 seconds on the same hardware.
Basically, you need to decide what you're going to use your system for (transaction processing versus ad-hoc queries), decide what performance level you actually need (i.e. how many transactions per second you're planning to handle). The performance of your system will also depend very much on how you write your queries and implement your indexes.
For most general purpose applications, SQL Server is fine.
|
|
|
|
|
Mike Dimmick wrote:
Frankly, SQL Server usually performs well enough on almost any hardware. Oracle really requires massive amounts of memory - 9i couldn't create a database on our test system with 192Mb (taking 6 hours and then the client crashed), while SQL Server 2000 creates a new database in about 5 seconds on the same hardware.
I looked at the URL you specified... SQL Server isn't doing so bad at all.. But I remember I read some performance reports from Oracle site. According to their report, SQL performance is just a small fraction of Oracle.. I didn't read into the details however.
|
|
|
|
|
Mike Dimmick wrote:
Frankly, SQL Server usually performs well enough on almost any hardware. Oracle really requires massive amounts of memory - 9i couldn't create a database on our test system with 192Mb (taking 6 hours and then the client crashed), while SQL Server 2000 creates a new database in about 5 seconds on the same hardware.
Not to mention, SQL Server is organized in such a way as to make it easy for "non-sql" people to use. With the different flavors (MSDE, Standard, Developer, and Enterprise), you can really get just what you need without the tons of overhead that Oracle requires.
Also, with Oracle, from around 7.0.4 and above, you basically need at least a week of classes just to figure out the whole Scott/Tiger thing and how to get the DB up and running. That does not include performance tuning, backups or how and why the data dictionary works.
Mike Dimmick wrote:
Basically, you need to decide what you're going to use your system for (transaction processing versus ad-hoc queries), decide what performance level you actually need (i.e. how many transactions per second you're planning to handle). The performance of your system will also depend very much on how you write your queries and implement your indexes.
We did a lot of performance testing on both Oracle 9i and Sql Server 2K. We found that when it came to data warehousing (large queries, mostly static indexes and a lot of tables!), Oracle was best. When it came to more transactional queries, Sql Server was top notch. Plus, given that .NET is tailored to Sql Server (go figure!), there is an added performance boost... 
Ok, enough of my rambling... 
Bill P.
Oakland, CA
-----BEGIN GEEK CODE BLOCK-----
Version: 3.21
GCM/MU/B dpu s--:-- a32 C++++$ ULH+++ P+++ L++ E+ W+++$ N++ o K? w++++$ O-- M V-- PS+ PE+ Y++ PGP++ t++@ 5++@ X++ R+@ tv b++ DI++ D+++>++++ G++ e++ h---- r+++ y++++
-----END GEEK CODE BLOCK-----
|
|
|
|
|
As others have said, it all depends on your available hardware and the type of DB you want to create. I have found SQL Server to be much easier to use (in general), easier to manage for most day-to-day taks, has better management tools and performs well on updates/deletes and good-enough on selects. Oracle has poorer management tools, but has many more configuration options readily accessible and performs better for querying, but unless configured to the Nth degree, has poor performance on inserts/updates/deletes.
If you know how to configure Oracle, I think you can squeeze better performance out of it on an apples-for-apples hardware comparison. Also, Oracle IMHO is more scalable to multiple servers, especially in geographically seperated situations.
Ultimately, I choose SQL Server because I just don't know how to configure all the options and SQL Server sets most things automatically and does a good enough job. About all you need to know specific to SQL server (beyond basic DB design) is how to use file groups. Also, the management tools for SQL Server are much better than the comporable Oracle tools.
|
|
|
|
|
Hi everyone,
I have not been in touch with SQL for a long time. I had to start using it again since a few weeks and it has been quite a struggle.
I have a question and I would be really grateful if someone can help me with it. Consider the following table structure:
ID TESTNUMBER STATE VALUE
1 100 1 0
2 100 1 1
3 100 2 1
4 100 2 0
5 101 1 1
6 101 1 0
7 101 2 1
Now, what I want to do is get the DISTINCZ count of TESTNUMBER, say where (State = 1, Value = 0) and (State = 2, Value = 1). So basically this should give me a count of 2 because TestNumber 100 has an entry for (State = 1, Value = 0) and (State = 2, Value = 1). Also, entry 101 has (State = 1, Value = 0) and (State = 2, Value = 1) entries. However, I am unable to write this using a query. Please help!
Thanks,
Sincerely,
Pankaj
Without struggle, there is no progress
|
|
|
|
|
Try this:
SELECT COUNT(*), state, value
FROM table
GROUP BY state, value
The statement above will group your result based on state and value, e.g.
Count State Value
2 1 0
2 1 1
2 2 1
1 2 0
If you need to add a condition, add HAVING state = 1 AND value = 0 (for example).
|
|
|
|
|
Hi,
I think I did not explain my question well. Let me explain the scenario. I have a code that has to generate all possible combinations of chosen States and Values.. Say for our example the combinations are (State = 1, 0) (State = 2, 1).
Now, I have to go in the database and find the TestNumber which has an entry for (State = 1, 0) AND (State = 2, 0) (which would only be TestNumber 100. If I had used an OR in my Select query it would return both 100 and 101. Also, if I use AND for the same field it would return me nothing)... So all combinations exist for this state. What SQL construct exists for performing such queries?
Thanks,
Pankaj
Without struggle, there is no progress
|
|
|
|
|
You can join your table to itself to do this. (I think this is what you are asking for...
select t1.testnum, count(t1.testnum)
from test t1
join test t2 on (t1.testnum = t2.testnum)
where t1.state = 2 and t1.value = 1
and t2.state = 2 and t2.value = 0
group by t1.testnum
this selects the testnum and a count from the table, joins itself based on the testnum and the where clause specifies where your first criteria matches on the "first" table and the second criteria matches on the "second" (your self join) table. Group it by testnum to see the count for each testnum that matches. Here are the results of that query...
Testnum count
100 1
If you change the t2.state = 2 to t2.state = 1, you get back
100 1
101 1
If this isn't what you are after, enlighten me some more on what you want.
|
|
|
|
|
Hi,
It had takem me a while but I figured that out. Thanks a lot for replying. I have a more complex one for you. I am really no good with SQL and unfortunately I have to work with it on this one 
Consider the following table again:
ID TESTNUMBER STATE VALUE
1 100 1 0
2 100 1 1
3 100 2 1
4 100 2 0
5 101 1 1
6 101 1 0
7 101 2 1
8 101 2 0
9 103 1 1
10 103 3 1
Now, I have to get all TESTNUMBER which have different sets of State and Values. So, this table will return 2 sets...
Set 1:
1 0
1 1
2 0
2 1
(Since, 100 and 101 share a common set)
Set 2:
1 1
3 1
(102 does have an identical set as 100 and 101. Common items are allowed but the sets must be identical (same elements))
Any suggestions on how to achieve this?
Thanks a lot,
Pankaj
Without struggle, there is no progress
|
|
|
|
|
Honestly, something that convoluted, I would just query the table and figure out the sets in my app side code. You're more than likely going to have to loop through the records, and I'd rather do that in code than with SQL Cursors.
|
|
|
|
|
Yeah, that is what I ended up doing. I am not a big fan of SQL 
Thanks,
Pankaj
Without struggle, there is no progress
|
|
|
|
|
ListObjectPermissions
what does the above mention statement do??
can i use this to get the access permissions of a user?
ranjani
|
|
|
|
|
ListUserPermissions
what does this method do..
does this tell me the user access perm for a particular database?
ranjani
|
|
|
|
|
Ranjani,
You post messages pretty often. Why don't you sign up as a user? Just wondering.
Basically, the way SQL Server permissions go, you have permissions as the result of being a member of a Windows group or role, right? (When I speak of permissions, I also mean that you can be denied access to something.) Then on top of these group-related permissions, you can be explicitly granted or denied extra permissions. These are usually granted/denied/revoked using the GRANT, REVOKE, and DENY statements in Transact-SQL, since most admin types use Transact-SQL to script this stuff up.
The ListUserPermissions method lists the permissions of the second type: special added-on permissions granted specially for a specific user. It won't include permissions that a user has through inclusion in a role, for instance, so be careful.
Also, you should realize that the Permission object can represent the fact that a user is denied access to something-- check out the Granted property.
Hope this helps.
Regards,
Jeff Varszegi
|
|
|
|
|
HI ,
I have made a procedure for sql server with visual c++ and i have test it on my computer and it is well but when i have move it to the server it could not work and it give me the following error
ODBC: Msg 0, Level 16, State 1
Cannot load the DLL sim.dll, or one of the DLLs it references. Reason: 126(The specified module could not be found.).
can any body help me
|
|
|
|
|
You've probably used some run-time library that you have on your computer but isn't on the server.
To check dependencies for a DLL, get hold of Dependency Walker[^].
If you have built a Debug build of your DLL, I recommend building a Release build and testing that on your computer before deploying that to the server.
|
|
|
|
|
I have build the release one and tested it on my local computer and it is working , and for information i did not use any external library just the standered one , and i still have the problem
|
|
|
|
|
If your DLL was built using Visual C++ .NET 2002 or 2003, and you opted to use the multithreaded DLL version of the C runtime library, there's a new version that may not be on the server.
For 2002, you need msvcr70.dll; for 2003 it's msvcr71.dll.
If you've used any C++ standard library features (for example, iostreams) you might also need msvcp7x.dll. If you used 'old' iostreams (declared in <iostream.h> rather than <iostream>) you'll also need msvci7x.dll.
Even if you only used Visual C++ 6.0 to compile, the server might not have all the required C runtime DLLs, although this is less likely. Windows NT 4.0 did not originally ship with msvcrt.dll version 6.0, but it's included with SQL Server 2000.
I still recommend using Dependency Walker to view the actual DLLs required by your DLL. Using it on the server will show what's missing.
The other possibilities I can think of are that you've specified the wrong DLL name - although I would expect that you'd noticed this - or that the DLL is in the wrong directory. You may be able to specify a full path - I haven't tried this.
|
|
|
|
|
Thaks for your help i have copy the msvcrt71.dll file to the server and it works .
Mhmoud Rawas
------------
Software Eng.
|
|
|
|
|
Is it possible to script a dts package the way one can script an sp? I was thinking that running the script through query analyser might help me discover exacly where the package is failing in copying the larger of our 3 db's to our disaster recovery server, but I haven't managed to find an option (other than saving as a VB file).
Thanks again
Look at the world about you and trust to your own convictions. - Ansel Adams
Meg's World - Blog
Photography - The product of my passion
|
|
|
|
|
Hi,
I've the following requirements:
1) Allow users to edit columns 2 & 3 of a DataGrid.
2) Prevent users from adding rows.
I can solve (1) by checking the event of moving cells, and then doing the following:
---------------------------------------------------------------
if (this.DataGrid1.CurrentCell.Columnnumber==0)
this.DataGrid1.ReadOnly=true;
else
this.DataGrid1.ReadOnly=false;
---------------------------------------------------------------
This works really neatly. I can't figure out how to solve problem (2) though.
Any ideas?
Pauwl
|
|
|
|
|
IIRC, it's a property on the underlying DataView: AllowNew.
|
|
|
|
|
Mike Dimmick wrote:
IIRC, it's a property on the underlying DataView: AllowNew.
Thanks Mike, this does the trick.
I somehow need to set this each time the control becomes active, but still, it works nicely.
thanks,
Pauwl
Pauwl
|
|
|
|
|
Could you please give me some website that can help me while migrating SQL to Oracle?
|
|
|
|
|
Oracle has a tool to migrate from SQL Server, I read about it in otn.oracle.com[^], but can't find the link.
Do some search in there.
Free your mind...
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
