There's something very reassuring about the knowledge that you have amassed a great wealth of data about your company. It's reminiscent of that old TV ad for an insurance company where the ecstatic customer is surrounded by an (anything but solid) cardboard castle which he taps happily and says "I've got the strength of my insurance policy around me" or some such guff. The feeling of security from mounds of data is great, but the practicality is shaky if it's put to the test. So often it proves impossible to shovel the data mountain into your brain in appropriate, bite-sized chunks.
Traditionally there have been two approaches to this problem. At the high end was the Executive Information System (EIS), where selected slices of the data were packaged up for the top brass in little more than a slide show. While intranets look likely to replace specialist programs and Visual Basic as the delivery mechanism, this end of the market remains very inflexible.
For analysts, on the other hand, you chomped off the appropriate data with a 4GL such as FOCUS and downloaded it into a spreadsheet to play tunes on the results. This is very flexible, but difficult to cope with.
Many managers want something with more flexibility than an EIS, but without the lack of guidance provided by the spreadsheet.
Spreadsheet vendors would have us believe that the latest generation of Excels and 1-2-3s are so powerful and easy to use that they can fulfil that middle role. There's an element of truth in this, especially if an IT professional can first massage the information and set up some jolly macros. After all, pivot tables and easy data access are key elements to slicing and dicing data. Yet a real business database can easily overwhelm a spreadsheet.
A lot of the problem is down to that hoary new chestnut, OLAP (OnLine Analytical Processing). Most of the world thinks this is a pretty novel idea, though APL programmers (all six of them) will point out that APL has been doing it since the 1970s - and in one line of code too - but that's a different story. OLAP turns the relational database on its head, multiplying up the dimensions of the data to provide an n-dimensional hypercube of information. Yet relational databases weren't invented for fun. The obsession with avoiding redundancy of data may be driven by making sure that it's only updated in one place, but it's also because OLAP produces big files. Very big files. The sort of thing that makes an Excel pivot table collapse in a giggling heap. Canadian firm Cognos thinks it has the answer with PowerPlay 5, a tool that can cope with more than 20 million records - try that in 1-2-3.
PowerPlay aims to make business intelligence tools as accessible as the familiar spreadsheet, so it's a good start that there's a proper InstallShield setup. It's tempting to overlook the fact that someone who's good with numbers isn't necessarily a computer expert. This is by preference a Windows 95 or NT application. It will run on Windows 3.1 using the Win32 libraries, but Cognos feels it is not at its best in this environment.
The first stage of applying PowerPlay is to employ Transformer. This takes your tables and combines them up to form the hypercubes of data Cognos calls PowerCubes. There may be some pre-pre-processing, because Transformer can only use a flat file, so any relational links have to be queried out to get the appropriate chunk of data. That flat file can come in a wide range of ASCII, spreadsheet and database formats, with all the desktop standards except Access covered. Note, though, that the import is a little fussy. For example, if you fail to name a range in Excel, it says it's the wrong file format. The administrator then takes the selected file and builds a cube from it.
This involves identifying the key indicators in your data and dragging them over to the measures box. You then set up a dimension map for your n-dimensional block of data. Hit the category generator button on the toolbar and off Transformer goes, looking for all the combinations of the dimensions you have established. You can now see and manipulate an effective graphical representation of the data structure before outputting the information to a cube. This process is quite time consuming. Just reading a 1,600 record file took around two minutes, during which there was no feedback whatsoever. Category identification inevitably takes a long time as Transformer is looking for every possible combination of your dimensions, and even producing the cube takes a while. What Transformer is crying out for is a wizard to lead the user through the complex process of identifying and structuring dimensions.
Assuming you have more than the basic edition of PowerPlay, Transformer is not limited to a simple cube. You can have multiple input files and cube groups that define sub-sets of the data. PowerPlay provides the facilities to limit access to different parts and levels of a cube, allowing the same dataset to serve multiple users. It is also possible to define more than one drill down route for more complex data where the direction is not obvious.
Once the cube is generated we move over to PowerPlay. This will be the first that many users see of the application. PowerPlay displays and manipulates the cube. Cognos uses the image of a Rubik's cube, and it's quite appropriate for the way an expert user can rattle around a data structure. PowerPlay has Reporter and Explorer modes, with a quick switch between the two from the toolbar. Explorer is the more strongly formatted of the two. It forces the information into a uniform format, showing two of the dimensions.
This allows for conventional drill down and up, and makes it easy to have summary headings and simple calculations.
Reporter is much more flexible as you can combine categories from different levels in the cube. The result can be quite confusing. For instance, if you are displaying a pie chart of your data and drill down a segment in Reporter mode you get everything you already had, plus the drilled-down data. In effect, the pie chart is now over 100%. This flexibility is good, but there's not enough to make it clear which mode you are in.
Pie charts are only the start. Power-Play displays tables plus the nine most common graph types at a click of the toolbar. This is great, but when it comes to manipulating the cube you hit the worst part of PowerPlay.
It comes down to two of the biggest factors in user interface design - consistency and visibility. The consistency problem is with practically every other Windows application. On-screen items like the neat little folders that represent dimensions simply don't respond the way you'd expect.
Single clicks, and double clicks, and drag and drop have to be used in very specific and generally unpredictable ways. On the plus side, though, there is good use of right mouse click menus, a useful fallback. Good visibility means you know where you are and how to get somewhere else at any time. But PowerPlay does very little to indicate how far you've drilled down, or how to get to another level. The worst example here is where there's an alternative drill down. There is no indication whatever that this is the case. You have to bring up a categories dialog and look at the tree structure to see if there are other options.
Once you've got the cut of information you want, reporting is straightforward, taking a sensible path between a screen dump and a true reporting package.
Since PowerPlay is both an OLE server and client you can use your favourite packages to put together a sensible report.
PowerPlay is very impressive, which it ought to be at this price. Though there are flaws in the user interface, they can be lived with by someone prepared to put the effort in. While it may not fulfil its stated aim, PowerPlay remains about the best way to fill the gap between EIS and spreadsheet.
PRODUCT AND CONTACT INFORMATION
PowerPlay comes in four versions.
- End User edition: u700 - supports one input file and one cube
- Administrator edition: u1,995 - adds the cube groups facility
- Database edition: u4,995 - provides facilities to store cubes in popular third-party database servers
- Server edition: u17,995 - provides facilities to create and update cubes on popular third-party database servers
Contact Cognos on 01344 486668. Web site hyperlink www.cognos.com
- Handles large amounts of data
- Flexible transformer tool
- Great for understanding your data
- User interface is surprisingly difficult.
Ceres, located in the asteroid belt, has a carbonaceous-rich upper crust, SwRI study claims
The spacecraft found traces of hydrogen and oxygen molecules, known as hydroxyls, embedded in the rocky surface of the asteroid
The skeleton was unearthed more than 20 years ago in South Africa
Moon's dark side is mountainous, rugged and never visible from the Earth