JDBC Drivers and Fastest Datatypes

[nicholas.wakefield]

Basically I want to read a table as fast as possible and rewrite on a
different db and was wondering if getBytes would be the fastest
approach. On some drivers I've made use of dbms specific extensions but
I was wondering if anyone had looked into this or had any better ideas.

I'm dealing with 100 million row tables and yes I could use a tool but
I'm still interested in peoples opinions.

 

[Joan]

Basically I want to read a table as fast as possible and
rewrite on a
different db and was wondering if getBytes would be the fastest
approach. On some drivers I've made use of dbms specific
extensions but
I was wondering if anyone had looked into this or had any
better ideas.

I'm dealing with 100 million row tables and yes I could use a
tool but
I'm still interested in peoples opinions.

Is it 100 (million row tables)
or (100 million) (row tables)
;-)

 

[nicholas.wakefield]

100 million rows.

 

[Malte]

Basically I want to read a table as fast as possible and rewrite on a
different db and was wondering if getBytes would be the fastest
approach. On some drivers I've made use of dbms specific extensions but
I was wondering if anyone had looked into this or had any better ideas.

I'm dealing with 100 million row tables and yes I could use a tool but
I'm still interested in peoples opinions.

I'd assume that the database you use can link to another database. In
this case I would look into using a stored procedure and have the
databases fight it out.

 

[Malte]

Basically I want to read a table as fast as possible and rewrite on a
different db and was wondering if getBytes would be the fastest
approach. On some drivers I've made use of dbms specific extensions but
I was wondering if anyone had looked into this or had any better ideas.

I'm dealing with 100 million row tables and yes I could use a tool but
I'm still interested in peoples opinions.

I'd assume that the database you use can link to another database. In
this case I would look into using a stored procedure and have the
databases fight it out.

 

[nicholas.wakefield]

The databases cannot link, the transfer has to be done via jdbc for
this test.

 

[steve]

The databases cannot link, the transfer has to be done via jdbc for
this test.

which data base, if it is oracle then you can setup a static link between 2
or more database tables, and you do not need java.


Steve

 

[Malte]

which data base, if it is oracle then you can setup a static link between 2
or more database tables, and you do not need java.


Steve

The OMjust said that a link is out of the question.

 

[steve]

The OMjust said that a link is out of the question.

why?


then use remote login.

 

[nicholas.wakefield]

You're killing me,

A static link is out of the question, the db could be anything and the
question is more about the JDBC api than anything?

JDBC is the only way it can be done in this case.

 

[steve]

You're killing me,

A static link is out of the question, the db could be anything and the
question is more about the JDBC api than anything?

JDBC is the only way it can be done in this case.

your the one that asked for better ideas.


how do you think you are going to maintain data integrity, by pulling 100
million rows off 1 database to put them back into another database.


if you are going to be silly about it then:

you are going to need a linux/unix machine with java & multiple >2ghz
processors.

and a fiber link between the machine & the 2 servers.

and perhaps the visigenic drivers, as they are the fastest.

 

[Frank]

Basically I want to read a table as fast as possible and rewrite on a
different db and was wondering if getBytes would be the fastest
approach. On some drivers I've made use of dbms specific extensions but
I was wondering if anyone had looked into this or had any better ideas.

I'm dealing with 100 million row tables and yes I could use a tool but
I'm still interested in peoples opinions.

Not JDBC-related, and probably fairly common knowledge, but make sure
you drop indexes from the destination database before a bulk insert like
this.

 

[nicholas.wakefield]

The better idea, was only within the realm of what is offered by JDBC.
As for the hardware, I think the e10K will suffice. Though I've never
tried the Visigenic drivers, so I'll look into that.