This video, Building a Data Mart with Pentaho Data Integration (
http://www.packtpub.com/building-a-d...egration/video), teaches how to apply the Kimball Dimensional Modeling technique with Pentaho Data Integration tool. At the beggining it disclaim it will not teach Dimensional Modeling or how to use PDI. Considering how sophisticated Kimball’s technique is, the content is very usefull for anybody how has to assembly such Dimenional Data Marts.
I liked it very much:
- The video has very high quality: 1920×1080 @ 25 fps and 44KHz audio stream.
- The video capture is very fluid, very stable, no skipping. The machine on which it was recorded is blazing fast, and paired along Ubuntu it just screams – no loading times, no lags, only smooth movements. You can see it is not an edited video because most of the time the mouse cursor does not do leap around the screen;
- The product is not only a video feed, it has a whole site with the video splitted in sections, make it easy and an ease to navigate, to play, stop and resume;
- The narration is very concise and sparingly (no hums and ehhs or silences.) You can see it has been edited to reach this quality;
- The product is very complete, with downloadable material;
- It is totally based on FOSS community editions softwares, so no extra cost at all to follow it (although it does use Pentaho Analyzer, but a free limited feature license one – very high added value because this is a very usefull tool;)
- It shows en passant how to use the EObjects Data Cleaner to achieve better data quality, which is a huge plus just by mere mentioning it;
- It is a very good course because it address not only the promised subject with detail, but it also teaches a lot on important aspects of data warehousing, like using columnar databases, caring about data quality, file and database versioning etc.;
- Most of the technical decisions are justified, which means you bask on the author experience.
I didn’t like it:
- The narrator has an accent with gets in the middle sometimes (I could never fully understand the second part titles), but you get accostumed to it. Also, I believe they’ve changed narrators a couple of times or maybe changed the pitch – the guy sounds a bit different at some chapters. No big deal;
- The sound has some volume and pitch wobling: sometimes (and quite frequently) it sounds like having skipped a frame or two and the voice goes into a higher pitch while decreasing the volume a bit. However, other than a little irritanting, it does not impair the hearing and you eventually also get accostumated to it;
- The use of JNDI connections: PDI has variables which are better fit to make connections portable than JNDI. However, it is a technical option you don’t need to follow and it does not negativelly affect the video or the content and its usability;
- On the Packt off-line video interface I missed very much some buttons to replay the video or go back a little, pause it and so on. When you download and play with your favorite media player everything is ok, tough.
Conclusion
I could go on and on listing things I liked about this video, but I’d became repetitive. On the other hand, the bad things are not really important. Besides the listed above, I could hardly find anything else.
The content is rich, detailed, carefully organized and laid out, very usefull and very practical. The video itself is very good, worth an in person course, technically very well done, and with a nice off-line interface. The negative points amounts to some quirks of the audio and the offline interface navigation buttons that I miss, but none of them are really important and do not stand in the way for an enjoyable experience.
All in all, Building a Data Mart with Pentaho Data Integration (
http://www.packtpub.com/building-a-d...egration/video) is a buy if you need to learn how to use PDI to design an ETL process for loading a Dimensional Model.
