Diff between server Job Parallel Job

rekha · Post by **rekha** » Fri Oct 14, 2005 2:16 pm

Hi

Can any one please tell me what is the main diff between server job and a parallel job . It will be great if one can give me a situation where one has to develop server job only.
Becuase right now in our environment we have only parallel jobs.
-rekha

logic · Post by **logic** » Fri Oct 14, 2005 2:34 pm

Hi rekha,
Search the forum , you will find lot of stuff Also read the following post,

Code: Select all

http://www.dsxchange.com/viewtopic.php?t=95903&highlight=search+posting

A quick answer to your question would be performance

kcbland · Post by **kcbland** » Fri Oct 14, 2005 2:45 pm

I'm not going to cover the differences, that you can read from the 4 years of posts talking about them. I'll answer why someone would build a server job in an environment where only parallel jobs were built.

Server jobs are extremely flexible, don't require a lot of thinking to build. The Server side of DataStage is the original tool. The vast majority of customers only have the Server side licensed. Server side jobs only have a few active stages, because those stages do so much work in an intuitive and graphical nature. The equivalent logic in a PX job could span many stages. The Server side job has the full feature of the easy to use DS BASIC language plus functions, you don't have to be writing custom buildops. The hash file stage is powerful in that it allows for reading, writing, and updating a reference dataset. There is no equivalent in PX. Server jobs have a fantastic interactive debugger that lets you set watch points and step thru the job execution at the link level.

The short answer is that Server jobs are really fast to create, debug, and test. If power/speed leveraging shared resources across multiple nodes is not your only consideration, then Server jobs do well.

This is my opinion, but it's funny how snobby some folks are about PX. DataStage Server had made Ardent/Ascential, and the majority of ETL side customers still only use Server. Even with the TX, PX, Profile, and Quality product additions, the vast number of users and implementations are all about Server. It's not dead yet.

prabu · Post by **prabu** » Fri Oct 14, 2005 2:47 pm

logic wrote:Hi rekha,
Search the forum , you will find lot of stuff Also read the following post,
Code: Select all
http://www.dsxchange.com/viewtopic.php?t=95903&highlight=search+posting
A quick answer to your question would be performance

ray.wurlod · Post by **ray.wurlod** » Fri Oct 14, 2005 6:32 pm

Server jobs are extremely flexible, don't require a lot of thinking to build.

But, like any endeavour, to be done well they do require some planning and preparation.

Have a source-to-target mapping document.

Have your metadata already in the Repository.

Parallel jobs require slightly more preparation and planning, in that you need to plan things like data type handling/conversion, data partitioning, and even choice of stage type - there are many more processing stage types than for server jobs.

lstsaur · Post by **lstsaur** » Fri Oct 14, 2005 10:17 pm

Ray,
It will cost about 675,000 for EE and 230,000 for Sever edition. Do you think is it worth to pay so much more for EE?

ArndW · Post by **ArndW** » Sat Oct 15, 2005 3:02 am

lstsaur,

yes, the price difference (whatever it may be) is definately worth it to many organizations.

ray.wurlod · Post by **ray.wurlod** » Sat Oct 15, 2005 3:28 am

Ray,
It will cost about 675,000 for EE and 230,000 for Sever edition. Do you think is it worth to pay so much more for EE?

How long is a piece of string?

If you want the right tool for your circumstances, you buy the correct tool for your circumstances.]

rekha · Post by **rekha** » Sat Oct 15, 2005 8:03 am

This link does not seem to work ....... can any body view this link

logic wrote:Hi rekha,
Search the forum , you will find lot of stuff Also read the following post,
Code: Select all
http://www.dsxchange.com/viewtopic.php?t=95903&highlight=search+posting
A quick answer to your question would be performance

rekha · Post by **rekha** » Sat Oct 15, 2005 8:04 am

This information was really helpful, thanks for your time

kcbland wrote:I'm not going to cover the differences, that you can read from the 4 years of posts talking about them. I'll answer why someone would build a server job in an environment where only parallel jobs were built.

Server jobs are extremely flexible, don't require a lot of thinking to build. The Server side of DataStage is the original tool. The vast majority of customers only have the Server side licensed. Server side jobs only have a few active stages, because those stages do so much work in an intuitive and graphical nature. The equivalent logic in a PX job could span many stages. The Server side job has the full feature of the easy to use DS BASIC language plus functions, you don't have to be writing custom buildops. The hash file stage is powerful in that it allows for reading, writing, and updating a reference dataset. There is no equivalent in PX. Server jobs have a fantastic interactive debugger that lets you set watch points and step thru the job execution at the link level.

The short answer is that Server jobs are really fast to create, debug, and test. If power/speed leveraging shared resources across multiple nodes is not your only consideration, then Server jobs do well.

This is my opinion, but it's funny how snobby some folks are about PX. DataStage Server had made Ardent/Ascential, and the majority of ETL side customers still only use Server. Even with the TX, PX, Profile, and Quality product additions, the vast number of users and implementations are all about Server. It's not dead yet.

chulett · Post by **chulett** » Sat Oct 15, 2005 8:13 am

rekha wrote:This link does not seem to work ....... can any body view this link

The link would have worked fine if it was posted properly as a link. The way it was done, it needed to be cut-and-pasted into the Address line of your browser to see.

The URL tags allow you to link to other postings rather easily, see? There is a trick to learn to embed them in another object, like a line of text. It would also have 'worked' if it was just pasted in the body of the message like this:

viewtopic.php?t=95903&highlight=search+posting

It is the code tags it was wrapped in that made it cease to function as a link. FYI.

rekha · Post by **rekha** » Sat Oct 15, 2005 8:54 am

I got it now .... the search in forum gave me old postings which are helpful

Thanks to all

chulett wrote:
rekha wrote:This link does not seem to work ....... can any body view this link
The link would have worked fine if it was posted properly as a link. The way it was done, it needed to be cut-and-pasted into the Address line of your browser to see.

The URL tags allow you to link to other postings rather easily, see? There is a trick to learn to embed them in another object, like a line of text. It would also have 'worked' if it was just pasted in the body of the message like this:

viewtopic.php?t=95903&highlight=search+posting

It is the code tags it was wrapped in that made it cease to function as a link. FYI.

kcbland · Post by **kcbland** » Sat Oct 15, 2005 9:51 am

lstsaur wrote:Ray,
It will cost about 675,000 for EE and 230,000 for Sever edition. Do you think is it worth to pay so much more for EE?

If what you need is a space shuttle, why look at airplanes? If your requirement is to process billions of rows of data, you have a large hardware budget allowing carte blanche on your configuration, and your team are intelligent and experienced data integrators, and money to spend on tools, why would you buy Server? Likewise, if your requirement is millions of rows of data, you have a modest budget allowing for a single server implementation, your team is average developers with a few superstars, and a small software budget, why would you buy EE?

The hardware, team, budgetary and processing requirements have to come together when making this decision. I would like a Ferrari F430, but even having such a fast car is a waste of money if the speed limit is 55. Seating for 2 doesn't work with my family of five. Now if the streets were rated like the Autobahn, and I was a single guy, that F430 starts looking like the better decision.

Ahhh... Ferraris...

prabu · Post by **prabu** » Mon Oct 17, 2005 3:06 pm

what about Parallel routines , which forces the DataStage folks to know C?? comments please

ray.wurlod · Post by **ray.wurlod** » Mon Oct 17, 2005 4:09 pm

Server routines force you to know DataStage BASIC.

And there are probably a lot more C programmers out there than there are DataStage/UniVerse BASIC programmers.

What's the point? If you're going to program, you need to know/learn the programming language of choice for the application.

DSXchange

Diff between server Job Parallel Job

Diff between server Job Parallel Job

Comments about Parallel routines