Thanks again for the input. You’re right about separating servers and tools, it’s just that with Oracle OLAP, a server decision (not to support MDX and XML/A) has a big impact on what tools can be used.

To take an example, I was at a customer site yesterday who very much wanted to use Oracle OLAP because of all the architectural reasons – single OLAP/RDBMS database, scalability, familiar Oracle environment and so on – but they chose not to in the end because they couldn’t find a query tool that still let them do all the financial analysis using ragged hierarchies, selecting dimension members based on their position in a hieraerchy and so on. I know D40 can do this, but the functionality of the tool just wasn’t enough for what they were looking for.

Whilst an SQL tool in the surface might have been enough (i.e. they can make selections based on attribute values and dimension member names), they just couldn’t work effectively with flattened, level-based hierarchies and accessing their cubes via SQL just wouldn’t cut it. For example, if their ragged hierarchy was seperated out into, say, 10 or 20 different levels, there’s no way of knowing in which level a dimension member may be placed, and so realistically they need a selector-style UI where they can pick dimension members based on a hierarchy. Now I’m willing to accept that this type of use case is far less prevalent than your normal “select dimension members based on attribute values” – I guess this is why Oracle went for SQL as the preferred OLAP query language – it does leave the more traditional OLAP users with few options, a situation made more difficult by the lack of any third-party OLAP tools support.

Anyway, as you say, it’d be good to see 11g OLAP and I sincerely hope it’s as good as the advance documentation suggests – I’m keen to see usage of the OLAP Option increase if only for personal reasons, as it’s something I’ve spent the last five or so years working on ;-) Let me know when webcasts and so on are scheduled for.

– regards, Mark

The OLAP Option is much more. It is a full featured calculation engine based on the dimensional model, it has the full dimensional model in the Oracle data dictionary and it continues to have a dimensional query API. The calculation model is fully capable of servicing financial reporting (about 1/3 of OLAP Option applications are in the financial reporting space). The calculation model was enhanced in 11g with calculation expressions that are dimensionally aware but are also based the basic grammar of SQL analytic functions. These expressions will appeal to a very broad audience of SQL literate developers and DBAs (unlike MDX).

The real point of the 11g OLAP release is that it brings the performance (both query and incremental update), analytic content and dimensional computation model to the much broader audience of SQL based business intelligence applications and application developers. 11g OLAP also centralizes the management of metadata (business model, calculations and the relational representation of the cube) in the Oracle data dictionary where it can be leveraged by a very wide variety of tools.

About tools …

The OLAP Option doesn’t need to rely on a specific set of tools built for it (with languages such as MDX). Any SQL based BI tool can query OLAP cubes and get the full value of OLAP performance and calculations. That’s a big world.

And, because dimensional concepts are revealed to SQL applications (e.g., value based hierarchies/parent child relationships, ancestors and descendants, etc.), they can easily issue SQL queries that leverage dimensional concepts. Also, it’s worth noting that SQL now has a PIVOT command with SELECT.

The point is that with 11g OLAP you don’t need tools that specifically support it – SQL based applications just work.

If we think about tools within Oracle, Oracle Business Intelligence Enterprise Edition is a perfect example. It can easily query OLAP 11g cubes, it can leverage hierarchical structures (e.g., WHERE product_parent = ‘FOO’, or WHERE product_brand = ‘TIDE’ AND product_level = ‘UPC’ and it can access any calculation within the cube. It can do this out of the box, without modification. That’s the whole point.

You need to take a look. We’ll be doing some public webcasts in the near future or I can give you a private demo.

Thanks for the feedback on the posting, of course the two of you more than anyone would know what is and what isn’t the strategic direction and so on, thanks for this. I guess my comment was based on the following observation:

- from what I’ve seen and heard of the functionality coming in 11g OLAP, it seems to me that the focus for this release is on “supercharging star schemas” – i.e. making OLAP a transparent performance booster for relational data warehousing. Now I see this really as a data warehousing feature rather than a traditional OLAP feature, and whilst it’s going to make the Oracle Database even more attractive as a DW platform, it doesn’t really do much to address the other OLAP usability issues – the lack of functionality in D4O, no support for MDX and XML/A (hence no true OLAP third-party tools) and so on.

- With Essbase though, you’ve got a whole bunch of OLAP apps, an MDX and XML/A interface out, more developed Excel tools and so on, a wide developer ecosystem, a fair amount of customers, and it’s cross-platform with no dependency on an Oracle Database, which seems to be flavour of the month at the moment.

It seems to me then that, in the scheme of things, Oracle OLAP will be positioned as a DW performance booster, whilst Essbase will be positioned for customers who wish to do the more traditional OLAP analysis – lots of ragged hierarchy reports, Balance sheet and P&L reporting, so on and so on, if only because Essbase has the front-end tools and I can’t see Oracle spending an awful lot of time enhancing D4O now. In particular, seeing as Essbase is starting to crop up in architecture diagrams for BIEE as a data source and data consumer, and as BIEE itself is the strategic BI direction, it seemed to me that Essbase, rather than Oracle OLAP, was going to get the attention in future.

That said – you’re obviously right, as you say both OLAP servers will be supported and maintained and developed into the future, and they’ve both got complementary roles – it’s just the message I’ve been hearing is that Oracle OLAP is going to be positioned more to enhance Oracle data warehouses, Essbase will be the recommended OLAP server for use in BIEE.

Anyway, as I said, thanks for the feedback, it’s certainly interesting to see how the products are panning out since the acquistions and release of new versions. I’m keen to download and have a play around with 11g when it’s finally released, I’ll be going through a similar exercise with Essbase once I get a bit of spare time.

– regards, Mark

The OLAP Option is Oracle’s strategic direction for OLAP in the data warehouse, with it’s primary role being to improve the query performance and add analytic content SQL based business intelligence applications (a far large market than traditional OLAP). Note that Oracle OLAP, particulary cube-organized materialized views, was a very prominate part of Chuck Rozwat’s session in the 11g launch event (see http://www.oracle.com/events/index.html). Doug Henschen of Intelligent Enterprise offered a nice write of up Oracle OLAP in his reveal of the 11g launch event (http://www.intelligententerprise.com/blog/archives/2007/07/oracles_11g_lau.html).

Similarly, Oracle OLAP (and Partitioning, Data Mining, Warehouse Builder) is strategic for Oracle’s analytic data warehouse. Oracle data warehousing is ranked #1 by every measurement (i.e. market share, scalability, etc.). Our goal is to make Oracle OLAP an integral part of *every* Oracle data warehousing implementation.

This is exemplified by the Oracle Database 11g release – where the embedded OLAP technology has been enhanced to make it leveraged more easily by the 250,000 Oracle Database customers and *all* business intelligence vendors running on Oracle. In 11g, Oracle OLAP cubes can be used as a storage mechanism for materialized views. To be clear, OLAP cubes are seamlessly embedded in the Orace Database materialized view infrastructure (for query rewrite and MV refresh). This can accelerate query performance for BI tools – without requiring the tools to make any changes to their applications.

But that is just about performance. Don’t forget the world class calculation engine that is central to the Oracle Database OLAP cubes. The cubes are automatically represented as a star schema in Oracle Database 11g. This delivers rich analytic content – quickly – to any SQL-based tool; advanced analytic calculations are exposed using very simple SQL. Star joins are optimized in the OLAP engine thanks to tight integration with the Oracle optimizer.

And, of course, key database functionality extends to Oracle OLAP. This includes: 1) Security: Object and data security, 2) Scalability: Real Application Clusters, 3) Availability: Backup-Recovery, Disaster Recovery, RAC, 4) Administration: Enterprise Manager and 5) Portability: All major platforms.

You will be hearing a lot more about Oracle OLAP in the coming weeks/months. Stay tuned.

OWB11gR1 will be out with the database, so I hear, there aren’t any major new features that will lead to delays. The big one will be OWB11gR2 – the IDE’s being changed to the JDeveloper/SQL Developer framework, mappings will be able to run real-time or batch, there’ll be Knowledge Module integration and so on – if any release will slip, it’ll be this one, there’s a fairly ambitious set of new features in there and I’d be suprised personally if it’s in our hands by the end of CY2008.

regards

Mark

;-)