We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.
Start a Discussion

Are OLAP Cubes at the End of Their Product Life Cycle?

Vote 0 Votes
Glen Rabie: Are OLAP cubes at the end of their product life cycle?

7 Replies

| Add a Reply
  • Interesting question! Part of me would like to say yes, but that definitely has to do with the fact that the companies I see are going about BI and analytics in a different way - by using data visualization and dashboards as an entry point to information access and distancing themselves from applications that are traditionally considered super user tools.

    As interactivity and user friendliness become more of an integral part of BI design I think that more companies will look for different ways to perform analytics and move away from OLAP use. However, I am sure that some businesses will maintain their OLAP cubes providing their super users get benefit from them. Unfortunately, at some point, those companies will be left behind the BI technology/advancement curve.

  • This is a question that needs to be answered with "it depends". The question is more - one of - if you roll up the cube information at run-time and present it as needed or do you create, update cubes in real-time and then the queries go against them when needed?

    If you are Wal-Mart, Target or a huge Supermarket Chain like Krogers or Safeway, it is silly and may be impossible to create these OLAP cubes at run-time to answer user queries. Given the number of SKUs and stores, if you want to analyze store-wise, SKUwise sales, doing it at run-time may not be possible.

    On the other hand, if you are Uncle Bob's Football Player Bobbleheads Outlet, you can form the cubes in real-time and answer questions. Even for Medium sized companies it may be possible to create cube-based reports at run-time like ROLAP based products like MicroStrategy.

    It depends upon two factors - number of dimensions and volume of transactions. If Both are high or either one is high, you may need to take a batch approach, create cubes, update them with transactions all the time. So that when you do the querying, you are querying cubes that have been updated already.

  • OLAP Cubes have a number of limitations that restrict their ability to support business decision making in today's fast-paced environment. Batch update, inflexibility, inability to apply analytics such as real-time scenario analysis, in addition to the usability issue for everyday people as Lyndsay has already pointed out. They may still have a place for strategic analysis by a super user or business analyst but businesses are increasingly looking for ways to improve execution and operational effectiveness.

    A new approach is required and the capabilities enabled by real-time (or right-time) Operational BI help everyone making daily operational decisions to make better decisions more quickly - and this delivers tremendous business value.

  • I think OLAP cubes have a few more years of life. To understand this, we need to look at the BI picture holistically. I have listed below two scenarios to make the point (Note: it may not be a comprehensive list):
    1) Real-Time BI with Real-time data - where the users are accessing real-time analytics as their BI servers are getting data from the operational environment as it is happening. Does every company need this? Absolutely not. 95% of users I have come across can live with some level of data latency ranging from 1-12 hrs.
    2) Real-Time BI with Batch data - where users are accessing data that is updated either once a day or as frequently as every hour (traditional DW environment). Users can use any of the dash-boarding / reporting tools to access this data (the presentation layer).

    OLAP cubes fall under the second category above. A small OLAP cube can contain millions of transactions pre-aggregated into meaningful business scenarios. OLAP cubes can be file-system based or DB based. The beauty of the file-system based OLAP cube is that you are not dependent on having an internet connectivity to access your data! Business travellers can "carry" a part of their DW with them in the form of an OLAP cube. OLAP cubes can be accessed in multiple ways - either through proprietary UIs or through the corporate standard dash-boarding / reporting tool.

    The main challengers to the OLAP cube technology are the new breed of in-memory BI tools with "associative" powers! While OLAP cubes depend on a pre-defined structure of data, the associative technology allows users to slice/dice the data in literally free-form (as long as some basic rules are met).

    Again, the footprint of the new technology tools are very small (but growing rapidly). So till then, its life as usual for OLAP cubes.

  • I think that OLAP cubes are at the end of their life cycle for a number of reasons.

    Firstly - the way in which BI is being delivered is changing as Lyndsay points out. Analysis is moving from the hands of experts to general business users. Their needs are probably more real time which is not as well supported as a cube but this is a questionable argument. I think it has more to do with the tools that deliver BI to these users that are less and less focused on cubes as a data source.

    Secondly - newer technologies such as in-memory databases, columnar databases and hardware appliances are addressing the original drivers for creating cubes in the first place - speed of analysis. The benefits of these solutions over a cube is less aggregation providing more flexibility of analysis.

    Thirdly - and not to be underestimated in the desire of technology companies to sell bright and new shiny objects. So in this sense OLAP will be replaced by the technologies listed in point two. Just take a look at Microsoft's own strategy with powerpivot for a glimpse at the future of the OLAP cube.

  • BI predominantly has been analyst driven and not business user driven. The reason being that most of the concepts – especially OLAP and the likes were derivatives of DBMS/RDBMS concepts and one should not forget Codd’s rules – has been ruling for a while – now that’s changing.

    As we see now, BI is slowly shifting towards a more business users driven approach - whether you call it operational, real-time, near real-time analytics - but the fact of the matter is that business users will have a completely different point of view – where the information lookup pattern is sharply top-down and system of records becomes less important as time goes by.

    So all in all it’s not just about OLAP but BI in general will have to face a much expected EOL and morph into a keyword driven, event aware, engagement based self learning entity – feeding the business users with the appropriate information whenever required – either lookup based or event based.

  • I, for one, still believe that the traditional OLAP cubes (aggregated, batch-updated etc.) satisfies certain analytical use cases for which there is no credible alternative. The classic use-case as Eash Iyer points above is the notion of 'disconnected analytics' - cubes with pre-aggregated information residing on laptops of people on the move.

    Also, BI technology practitioners (including me) wrongly assume that the business users would want to analyze data all day, if given the right technology. In an ideal scenario, the business executive wants to get the right insight in the shortest time possible and then spend more time in taking the decision and implementing it. In other words, the marketing manager provided with the right OLAP cube with marketing & related information, is more effective as compared to a situation where she is provided all the enterprise data and then asked to go 'figure out' whatever might be useful for her.

    So my verdict - Hybrid OLAP in the broadest sense will continue to help executives analyze their business.

Add a Reply

Recently Commented On

Monthly Archives