Hi All,
Can any one kindly provide what are the Hardware requirements for Server and PX Jobs .
I request what is the recommended Hardware requirement for Installing Server and PX.
Thanks in Advance,
Lucky
Hardware requirements
Moderators: chulett, rschirm, roy
Hello Lucky,
the hardware requirements for DataStage are not a fixed value and depend very heavily upon the expected usage.
The minimum to install is the disk space requirements of the server. DataStage itself reserves some shared memory segments and starts at least one background process (that is the minimum, on UNIX). It therefore doesn't need a much in the way of resources to just install.
Developers just designing jobs are also not resource intensive, most of the work is done on the client PC system. So even a minimally configured slow system will support multiple developers just writing jobs.
Once they start executing jobs in a development role, or if you are executing jobs in test or production it becomes a whole different matter. Depending upon how the jobs are designed, each execution will start off one or more processes, and these will fight hard to get as many resources as they can get.
If your database of choice is on the same machine then both DataStage and the DB will attempt to hog I/O, CPU and Memory and performance can drop significantly. If the database is on a remote system the network bandwidth is going to be used heavily moving data back and forth.
Each ETL project has a different mix of runtime requirements. Some do heavy data manipulation and transformations, these types of installations really need a large amount of CPU and tend to leave the I/O subsystems at low usage levels. Others need to transfer and split large amounts of data so have low %age CPU usage but bring network or disk I/O to 100% capacity. Both of these two general types of projects will want as much physical memory as possible to speed things up.
DataStage will "limp" along with a minimal hardware configuration and it will function, but very slowly. More hardware (up to a point, of course) will make DataStage run faster. Without knowing the usage profile it is very difficult to make any concrete suggestions on sizing.
the hardware requirements for DataStage are not a fixed value and depend very heavily upon the expected usage.
The minimum to install is the disk space requirements of the server. DataStage itself reserves some shared memory segments and starts at least one background process (that is the minimum, on UNIX). It therefore doesn't need a much in the way of resources to just install.
Developers just designing jobs are also not resource intensive, most of the work is done on the client PC system. So even a minimally configured slow system will support multiple developers just writing jobs.
Once they start executing jobs in a development role, or if you are executing jobs in test or production it becomes a whole different matter. Depending upon how the jobs are designed, each execution will start off one or more processes, and these will fight hard to get as many resources as they can get.
If your database of choice is on the same machine then both DataStage and the DB will attempt to hog I/O, CPU and Memory and performance can drop significantly. If the database is on a remote system the network bandwidth is going to be used heavily moving data back and forth.
Each ETL project has a different mix of runtime requirements. Some do heavy data manipulation and transformations, these types of installations really need a large amount of CPU and tend to leave the I/O subsystems at low usage levels. Others need to transfer and split large amounts of data so have low %age CPU usage but bring network or disk I/O to 100% capacity. Both of these two general types of projects will want as much physical memory as possible to speed things up.
DataStage will "limp" along with a minimal hardware configuration and it will function, but very slowly. More hardware (up to a point, of course) will make DataStage run faster. Without knowing the usage profile it is very difficult to make any concrete suggestions on sizing.
</a>-
ray.wurlod
- Participant
- Posts: 54607
- Joined: Wed Oct 23, 2002 10:52 pm
- Location: Sydney, Australia
- Contact:
In a word, "more".
The vendor publishes recommended and minimum specifications for different platforms. More is better, because - over time - the volume of data you process grows and grows.
In general it is cheaper to over-configure now than to increment later.
The vendor publishes recommended and minimum specifications for different platforms. More is better, because - over time - the volume of data you process grows and grows.
In general it is cheaper to over-configure now than to increment later.
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.