Greetings,
We are currently using DataStage 5.2 on Sun Solaris. We are planning to migrate to DataStage 6 next year Feb.
1. What would be the amount of work involved in the migration process.
2. Currently we are using ORAOCI8 to connect to the oracle 9i database. And since we have Oracle 9i plugin what would be the amount of work in migrating the plugin stages ?
3. Can I use the existing project and still incorporate the latest features (Like interprocess, Link Partitioner/Collector etc) ?
Please let me know your thougyts on this.
Tx,
Badhri ...
www.cognizant.com
DataStage V5.2 to DataStage V6.0
Moderators: chulett, rschirm, roy
1. Nothing more than any other upgrade, and certainly less than a 4.x to 5.x migration. I'm still 'testing' mine in my spare time, but I haven't found anything I needed to do other than run the upgrade script. Yet.
2. Probably not worth it. Start using it in new jobs and *maybe* replace them when working on old jobs, but if there's no real benefit it may not be worth the effort. From the readme and the appearance of the stage, it's not even clear if there is even a difference yet.
3. Sure!
-craig
2. Probably not worth it. Start using it in new jobs and *maybe* replace them when working on old jobs, but if there's no real benefit it may not be worth the effort. From the readme and the appearance of the stage, it's not even clear if there is even a difference yet.
3. Sure!
-craig
Thanks Craig,
Couple of Questions
.
1. Is it not going to be a BIG rework from migrating the mappings to use the Interprocess, Link Partitioner/Collectorfeatures. I heard that they provide you with great performance benifits ???
2. What are the notable advantages of using Oracle 9i plug in over 8i.
Badhri ...
Couple of Questions
.1. Is it not going to be a BIG rework from migrating the mappings to use the Interprocess, Link Partitioner/Collectorfeatures. I heard that they provide you with great performance benifits ???
2. What are the notable advantages of using Oracle 9i plug in over 8i.
Badhri ...
1. Not sure as I really haven't looked into it yet, but I'm guessing you are correct. The 'link' stages shouldn't take alot of effort to work into any new jobs, they seem to simplify some functionality that people have been looking for (and coding around) for a long time. The 'interprocess' stages, on the other hand, will require (IMHO) someone sitting down and rethinking how they design jobs now and how to really take advantage of the new capabilities. I haven't taken the time to do much pondering along those lines.
2.At this point, I'm not sure there is *any* difference between the two plugin stages. All I can suggest is, if you are running 9i, then start using the 9i plugin... but again I wouldn't suggest going back and changing to it in all of your old jobs. I'll call support and see what they say on the issue and let you know.
2.At this point, I'm not sure there is *any* difference between the two plugin stages. All I can suggest is, if you are running 9i, then start using the 9i plugin... but again I wouldn't suggest going back and changing to it in all of your old jobs. I'll call support and see what they say on the issue and let you know.
DataStage 6 upgrade has been worth it for us. There are some immediate benefits we were able to take advantage of.
We have been using in-process and inter-process row buffering out of the box to improve job performance. You can turn this feature on at the job level (recommended) or the project level. We have seen good improvement when applied to a mulitude of our jobs(our project is 500 jobs). You have to understand how they work differently. We have not code IPC or partitioner/collector stages yet. They may be useful in the future.
I have detected 2 bugs concerning Inter-process that Ascential is looking in to. The first is where a job with a Sort stage blows up when using Inter-Process row buffering and the second is when a job reads/writes to the same has tables. There appears to be an out-of-sync scenario. If you start using these features be sure to do before/after file compares.
Other improvements are in-memory cached hash tables, better looking GUI, and Parallel Extender integration. DataStage 390 had some new features we were looking forward to like better messaging and end-of-data indicator.
MetaStage 6 has improved "process" metadata collection and reliability as well as SendToDB (sends metadata to relational target) and the ability to apply XSLT transformations to XML metadata output.
The deeper you get, the more you find.
Kevin Janes
We have been using in-process and inter-process row buffering out of the box to improve job performance. You can turn this feature on at the job level (recommended) or the project level. We have seen good improvement when applied to a mulitude of our jobs(our project is 500 jobs). You have to understand how they work differently. We have not code IPC or partitioner/collector stages yet. They may be useful in the future.
I have detected 2 bugs concerning Inter-process that Ascential is looking in to. The first is where a job with a Sort stage blows up when using Inter-Process row buffering and the second is when a job reads/writes to the same has tables. There appears to be an out-of-sync scenario. If you start using these features be sure to do before/after file compares.
Other improvements are in-memory cached hash tables, better looking GUI, and Parallel Extender integration. DataStage 390 had some new features we were looking forward to like better messaging and end-of-data indicator.
MetaStage 6 has improved "process" metadata collection and reliability as well as SendToDB (sends metadata to relational target) and the ability to apply XSLT transformations to XML metadata output.
The deeper you get, the more you find.
Kevin Janes
Thanks, good to know. I checked with support about the two plugins and primarily the difference now is in the Client Software they support. Check on your DataStage cdrom under 'DataStage Packages' for the Oracle8i.pdf and Oracle9i.pdf files that explain this in great detail. Also note you can't mix 8i and 9i plugins in the same job.
-craig
-craig