Hi Dave. This will be a parallel response since Arbor and Hyperion are really different animals.
No not a massive list of criticism or comments, but a few.
RE: Arbor.
Oracle has a very difficult time making good use of any acquisition. They toasted Datalogix too. Not surprising at all to see Express lose often to Hyperion. Also not surprising to see the lack of Oracle in Hyperion's strategic alliances list.
I was referring to just the database capabilities of Oracle and DB2, not Express. As a logical data modeller, the whole multi-dimensional thing to me is overblown. It is possible to create very usable multi-dimensional models in standard SQL databases without a product like Arbor. In fact I have one where I am currently consulting. Four dimensions. I don't need no stinkin' multi-dimensional product to accomplish the same results I can get with good 'ole Oracle, Informix, DB2, maybe for small datasets even SQL Server. As for transform logic and interface handling. It is a lot simpler to use SQL and go direct from operational datasources to aggregates in the same environment than to batch stuff up and move it to the 'warehouse' or 'data mart'. If your operational systems are on an RDBMS, there is NO reason to put your dimensional data somewhere else is my point.
BTW, most of the products that do transforms lack what I would consider good batch balancing controls. The result is that you end up with a bunch of folks who do nothing but spend their time reconciling your dimensions to your operational data sources. If you have lots of countries, or business units, this is a big mess.
This is why I dislike the terms OLAP and Data Warehouse. A competent data modeller doesn't need any specialized software to do either. Just a good RDBMS and a high quality SQL implementation.
RE: Hyperion
One of my clients, GE is a very large Hyperion user. To me it has always seemed like spreadsheets on steroids. They love/hate it. It adds several days to their fiscal close every month while they reformat, verify, cross check and lord knows what else, the data from their operational GL systems. These interfaces take time. They are all batch, not real time. It is unnecessary. That is why the apps vendors are incented by their clients to provide native tools that don't require all this data movement. See above. I should be able to produce my close, do my budgeting and all my fiscal reporting directly from my operational systems -- if I have the horsepower on the operational system. That is the biggest issue. We can get into this in more detail some other time.
RE: Y2K -- I'm sure these software products are Y2K compliant. That isn't at all the issue. These products are application development environments in their own right. The provide a means to transform and store data in different formats via their own programming languages, whether SQL or something else. This is USER written code and it contains date logic.
My point is that the financial managers in business units that use this stuff ARE NOT testing their in house written transform and aggregation code for Y2K bugs, and there ARE Y2K bugs in their stuff.
It is like having complex Microsoft Excel Workbooks containing macros that perform some date work. The accountant who wrote them doesn't have a copy on a nearby PC with a post Y2K date for testing purposes. He not only doesn't have the facilities for testing, but doesn't have the skill set to do it in many cases.
Exact same thing is true of Hyperion and Arbor centric applications. I have yet to see a company with a plan to test these applications. And MOST of them are using this stuff to do their fiscals and annual reports.
We'll a few of them will be delayed, and a few of them will have inaccuracies that will go undetected until after the numbers become public. 'So what if the Canadian division numbers accidentally contained all of the last three years data because the inclusion range became 1/1/1900 to 12/01/1999.' See my point. Now if this was just one company, no big deal. Even Cendant will recover. But is isn't. It is ALL the big companies using products like Arbor and Hyperion for the complex job of producing their fiscal reports.
There are entire classes of important applications that are being missed when it comes to Y2K testing.
Sorry to be so windy.
Cheers
ttf
P.S. I don't like Express either. |
