PDA

View Full Version : How to delete a record in HBase table using Pentaho



varaprasad2
09-20-2012, 10:39 AM
Hi,

I have to delete a record/records in a HBase table using Pentaho.
Can anyone please suggest me the solution on this?

Mark
09-20-2012, 10:34 PM
You could use the Javascript or User defined java class PDI step in order to use the HBase API to connect and then execute a "Delete".

Cheers,
Mark.

ram_sivakumar
09-21-2012, 06:46 AM
Can u pls provide a sample script

Regards,
Rams

joniloky
10-17-2012, 03:31 PM
You could use the Javascript or User defined java class PDI step in order to use the HBase API to connect and then execute a "Delete".

Cheers,
Mark.

Is it just me? Most of the times I look up how to do something with PDI the answer is "code it in JAVA or JavaScript". I'm working with PDI to avoid writing code. That's why I give up having low level control over the ETL, so that I can put together a workflow using a simple graphic UI that non-technical people can understand and even modify. Maybe I got the wrong tool for the job.

Sorry guys, there should be ... tag for comments. I just wanted to put it out there and get your opinion.

jdixon
10-17-2012, 03:57 PM
You raise an interesting point. A number of existing PDI steps started out as a custom-coded Java or Javascript step that were sufficiently useful to be converted into full steps.

Most software is a work in progress, so new capabilities are added and existing ones are enhanced. I think we should consider the alternatives:
1) Design every conceivable option and feature up-front and don't release it until every one is implemented. This results in monolithic software that never gets released.
2) Deliver the software in incremental releases using a closed/un-pluggable/proprietary model. Using this approach the answer to many questions is 'no, you can't do that, you cannot extend the software and you have to wait for us to code it for you, if ever'.
3) Deliver the software incrementally using an open model. Using this approach the worst possible answer is 'it doesn't do that right now, but you can extend the system to do that'

So, I agree that deleting a record in HBase is a good feature. PDI doesn't do it today, but you can add that feature and contribute it, or you can work around it with one of the existing code/script steps. Either way I think it's better than "no you can't".

James

joniloky
10-17-2012, 04:17 PM
You raise an interesting point. A number of existing PDI steps started out as a custom-coded Java or Javascript step that were sufficiently useful to be converted into full steps.

Most software is a work in progress, so new capabilities are added and existing ones are enhanced. I think we should consider the alternatives:
1) Design every conceivable option and feature up-front and don't release it until every one is implemented. This results in monolithic software that never gets released.
2) Deliver the software in incremental releases using a closed/un-pluggable/proprietary model. Using this approach the answer to many questions is 'no, you can't do that, you cannot extend the software and you have to wait for us to code it for you, if ever'.
3) Deliver the software incrementally using an open model. Using this approach the worst possible answer is 'it doesn't do that right now, but you can extend the system to do that'

So, I agree that deleting a record in HBase is a good feature. PDI doesn't do it today, but you can add that feature and contribute it, or you can work around it with one of the existing code/script steps. Either way I think it's better than "no you can't".

James

I would agree with you 100% if not be it for the specific example we are discussing. Deleting a record from HBase seems like some of the most basic functionality that could be provided when integrating the technology into PDI. Every time you create persistent entities in a system the most basic functionality you are going to need is CRUD (Create, Read, Update, Delete).

This kind of basic functionality holes keep popping up. And that's what's frustrating.