Main Differences Between Server and Parallel Jobs?

[skumar]

Hi All,

Can Some Body tell me what are the Main Differences Between Server and Parallel Jobs.I know that one difference is in server we can write the coding in BASIC and in parallel we will write in C and also performance wise parallel will be faster.So plz let me know what are the Major other differences?


Thanks in Advance.

Regards,
skumar.

[Kirtikumar]

Just do search on 'Difference server and parallel'.

You will get so many posts and discussions on this topic.

[ray.wurlod]

The assertion "performance wise parallel will be faster" is not necessarily true, particularly for small jobs.

Why not enrol on the Server to Parallel Transition Lab offered by DCS (the company that hosts this site)? This class spends three days examining the differences.

[Prashantoncyber]

The Basic Difference is in that the Server Engine is based on Universe and Parallel Engine is on Orchestrate.

The Parallel jobs are better in performance because partitioning is built into each stage and it can use the configuration file which defines how system resources can be used effectively.

[ray.wurlod]

Startup time is substantially greater for parallel jobs than for server jobs. Therefore to do a small run, a server job is probably going to start, and therefore finish, faster. And to process a single row (such as obtaining the next surrogate key value), a parallel job is definitely disadvantageous.

Each tool has its place. Any client who forbids the use of server jobs is doing so from a position of ignorance. There has been some fear, uncertainty and doubt (spread mainly by sales representatives who get a larger commission from enterprise edition) that server jobs will "go away"; this has been explicitly refuted by senior IBM staff at last year's event.

I believe that, out of the box, the available set of transformation capabilities is richer in the server job environment than in the parallel job environment.

[vijaya]

Hi skumar,

Here are the some of the differances

Here are some of the differences I could come up with:

* Both Server and Px (parallel, Enterprise) might share the Ascential name but they evolved quite separately from each other, so it is not surprising that they handle certain things differently.

* Parallel and Server are COMPLETELY separate ETL tools under the covers . They share the terminology but do not share the same technology.

* Parallel writes C code that uses a C compiler and uses a custom script process (OSH) that is interpreted at runtime and is built for high-performance high-volume data loading . Server writes a version of BASIC programs using an internal compiler and executes via the DataStage shell (dssh).

* Server jobs are good for smaller volumes as they are less fussy about metadata. Depending on the design most server jobs process one row at a time.

* The obvious incentive for going parallel is data volume. Parallel jobs can remove bottlenecks and run across multiple nodes in a cluster for almost unlimited scalability. At this point parallel jobs become the faster and easier option.

* Understanding errors is much easier in Server than it is in PX. Parallel job problems seem to be more numerous and harder to debug than server job problems.

* In Px Buildops (user-defined routines) are a bit harder to code than server routines. You need to develop external C routines in Px where as user-defined routines can be directly coded in DS environment using DataStage BASIC.

* Database errors are harder to trap in PX.

* Debug mode is available with server jobs where you need use Debug stages such as Peek in parallel.

Regards,
vijaya

[lstsaur]

The Enterprise Edition is way much more expensive than the Server Edition. It's also difficult to find experienced PX developers.