DBMS Benchmark code ? Who's fastest ?

DBMS Benchmark code ? Who's fastest ?

Post by Orri Erlin » Fri, 12 Apr 1996 04:00:00



Does anybody have implementations of the TPC or other DBMS
benchmarks on the ODBC API ?

An implementation of those benchmarks on Oracles dynamic SQL
API would also be of interest.

The TPC documents do have sample implementations in C and
embedded SQL but that is not likely to be the
code actually used in performing the measurements.

We have a new ORDBMS called Kubl that we would like to
benchmark against the relational  performance
leaders.  You are welcome to check out www.infosto.fi/kubl for
information on the new contender.

--
Orri Erling

 
 
 

DBMS Benchmark code ? Who's fastest ?

Post by Akmal B Chaudhr » Thu, 18 Apr 1996 04:00:00



Quote:>=20
>=20
>=20
> Does anybody have implementations of the TPC or other DBMS=20
> benchmarks on the ODBC API ?
>=20
> An implementation of those benchmarks on Oracle=92s dynamic SQL=20
> API would also be of interest.=20
>=20
> The TPC documents do have sample implementations in C and=20
> embedded SQL but that is not likely to be the=20
> code actually used in performing the measurements.
>=20
> We have a new ORDBMS called Kubl that we would like to=20
> benchmark against the relational  performance=20
> leaders.  You are welcome to check out www.infosto.fi/kubl for=20
> information on the new contender.=20
>=20
>=20
> --
> Orri Erling
>=20
>=20
>=20

Orri,

There was some empirical work done here at City U a few years ago which
showed that the TPC Benchmarks are not representative of OLTP. The City
Benchmark is based on the observations of 40 million transactions (160
actually collected), 4800 discrete applications and 5000 Relational
tables. The work was part of the PhD Thesis of Dr. Wagdy Youssef. I can
send you the benchmark source code (which is commented) + a couple of
papers which describe its use in performance measurement. There is no
documentation per se, but if you understand database performance, then you
will have no trouble using it.=20

Regards,

=09Akmal.

     _  ___________________________________
    //)|                                   |
   / / | http://www.soi.city.ac.uk/~akmal/ |

(((\ \>|_/()_______________________________|
(\\\\ \_/ /
 \       /
  \    _/ "If I hear the phrase 'everything is an object' once more,
  /   /    I think I will scream."
 /___/                                     - Mike Stonebraker (1988)

 
 
 

1. ADO thinks DBMS is done, but DBMS ain't done

I have 2 Recordsets on 2 different applications communicating with the same
database

-=The first Recordset is opened with=-
m_pGlobal_RS->Open(adCmd, m_pConnection.GetInterfacePtr(), adOpenKeyset,
adLockPessimistic, adCmdText);

-=The second Recordset is opened with=-
local_RS->Open(adCmd, m_pConnection.GetInterfacePtr(), adOpenKeyset,
adLockReadOnly, adCmdText);

* adCmd is a simple SELECT statement that is the SAME for both recordsets.

The code order is roughly this

Open m_pGlobal_RS
...
...
Change fields on m_pGlobal_RS
...
...
MoveNext on all rows of m_pGlobal_RS until EOF then MoveFirst  // since
pessimistic every row op sends update
..
..
I then fire an event to another application which opens local_RS.

It is random whether the data opened by the second RS is the pre-update data
or the post-update data. How can this be? I opened the connection
with the default  synchronous option, isn't this asynchronous behavior?

Thanks for the help,
-=Adog

2. COB EDITOR EXTENSION

3. DBMS Measurement or Benchmark

4. SQLBase Database

5. DBMS Performance Benchmarks

6. Technical Support Specialists *LONDON*

7. DBMS benchmarks

8. ** Load image from Table to DBImage so slow...........????

9. Hypermodel/nonstandard DBMS benchmark

10. Looking for a DBMS benchmark

11. Where can I get benchmark test between DBMS?

12. Need benchmarks for LARGE db's on Unix (10's of Gig)