 |
29 June 2003 |
There's a thread going on at OTN about the Analytic Workspace Support in OWB. The original poster asks whether OWB can actually build and populate an AW, and when the required 9.2.0.3 database patch is coming out. Shauna O'Boyle has responded and there's some useful pointers and updates which I'm reproducing.
- The Oracle 9i 9.2.0.3 patch, which the Analytic Workspace Manager and OWB need to be able to build and populate Analytic Workspaces, is now due July 2003, although the specific date hasn't been confirmed.
- If you want to run BI Beans using a cube built using OWB, at present it has to be a ROLAP cube, at least until the 9.2.0.3 patch comes out.
- To deploy a ROLAP cube, you deploy the facts and dimensions as normal using OWB 9.0.4., then use the metadata bridge to create the CWM1 metadata.
- When the 9.2.0.3 patch comes out, this bridge will then also be able to create the metadata for a MOLAP cube in the Analytic Workspace.
- If you are using the metadata bridge, you should apply the OWB 9.0.4.10 patch which has a number of improvements and bug fixes that will help deploying to OLAP.
In a separate thread, it appears that the next version of OWB will be OWB 9.2. It's due to be released in mid-July 2003. OWB 9.0.4 was quite a change from previous versions and i've had mixed feedback from colleagues about some of the new deployment and debugging features, although it must be said that it's good to get the SAP integrator back. It'll be interesting to see what OWB 9.2 delivers.
10:40:35 AM 
|
|
"Is Oracle about to get Netscaped?" asks CNN Money.
"Not only is the Redwood City, Calif.-based company still under pressure from IBM (IBM: Research, Estimates) in the corporate market, but it now faces serious competition from Microsoft (MSFT: Research, Estimates) on the lower end of the market and in the Windows space. Though Microsoft's database product, SQL Server, has been around since the mid-1990s, in the past three years it has improved dramatically, and last year Microsoft for the first time became the top vendor in the Windows-based database market.
Microsoft's tactics in the sector -- enter with low-price, "good enough" technology and nibble away market share from entrenched rivals slowly but surely -- sound familiar to anyone who followed the browser battles of the late 1990s. So, is Oracle in danger of getting Netscaped?"
10:40:21 AM
|
|
There's a good online guide to implementing type 2 SCDs using OWB on OTN. We've come up with a number of manual ways of doing this for a couple of clients using OWB and some custom procedures; this seems like another approach that could be useful.
10:40:11 AM
|
|
"Six years ago, Larry Ellison predicted that his newest product would rewrite the rules of corporate IT. We're still waiting."
10:40:01 AM
|
|
There's a good article over at Wired about the problems facing Sun Microsystems. They're finding their market eroded on the one hand by Microsoft, with Windows Server 2003 appealing to the Microsoft market, and Linux on the other, offering the performance and Unix environment their high-end customers want.
Sun have let the Sparc platform drift these past few years and Scott McNealy's got too caught up in a personal vendetta with Bill Gates. Sun is Solaris, and Solaris only makes sense on Sparc, and there's just no reason now to go with that platform when Intel hardware is so cheap and powerful.
10:39:42 AM
|
|
From cnet.com: Oracle vs. PeopleSoft: The Microsoft connection. "The headlines from the last week shouting about the rivalry between Larry Ellison and Craig Conway overshadowed the real subtext to the struggle for PeopleSoft: the coming competition between Oracle and Microsoft for who's going to be No. 1 with enterprise customers."
10:39:30 AM
|
|
I recently got hold of a HP h5450 iPAQ with built in bluetooth and WIFI, to work alongside a Nokia 6310i to give me 'always on' internet connectivity using GPRS. It was fun and games setting it up, but I eventually got it all working and i'm pretty pleased with the result, such that often when I visit client sites now I leave the laptop behind and just take my PDA and phone.
I've had the chance to try out both bluetooth and WIFI, and come to the conclusion that they're both good, in their own ways. WIFI is good around the house, and when I manage to find public hotspots (such as at Gatwick Airport North Terminal and the Shelbourne Hotel in Dublin, where you pay around £10 per hour, and at the Great Southern Hotel in Dublin Airport, where WIFI access is free), and bluetooth is great for connecting the PDA and phone such that I can just put the phone in my bag and forget about it. Different technologies to solve different problems.
However, Bob Frankston has written an article in which he declares that Bluetooth has failed. The article states that despite the fact it is wireless, it still has all of the limitations of wires. The premise of his argument is that as Bluetooth doesn't use real internet protocols, it is an evolutionary 'dead end' and in addition it still suffers from all the drawbacks of physical wires (proximity limitations, requirement for specific protocols/adapters, and so on).
I think he's missing the point. Bluetooth was never meant to be WIFI. It's just a replacement for wires and it's meant to be a simple protocol. It's not intended to be a networking protocol and functionally it's one up from infrared, one down from WIFI. As a poster put it on a subsequent Slashdot discussion, "As already mentioned, Bluetooth is not intended to be a networking technology. It is one up from I-R and one down from Wi-Fi. Its one up to I-R because it allows simple devices, close together, to communicate together, simply, and not need to be in line of sight. It is one down from Wi-Fi because there is no need to use something this heavy duty to transfer simple data, doing so would simply be cost restrictive and over kill - this would be akin to using 4 ton truck for moving a box's worth of paper in your office. People who understand Bluetooth are using it for things like wireless keyboards, mice and synching PDAs and mobile phone to PCs. Printers are a special case, since in most cases you would want to use Wi-Fi, but by using Bluetooth you allow a simple PDA to print out a document - I suppose printers are pushing Bluetooth to the limits of what it was designed for."
It's also worth bearing in mind where Bob Franskton is coming from. He's an employee of Microsoft who is an advocate of an alternate home communication technology, Home PNA that's been backed by Microsoft, in contrast to their lukewarm to non-existent support of Bluetooth, a European technology that Microsoft can't control.
Bluetooth works for me as a way of elminating wires, keeping communication lightweight and working universally between devices with little setup needed. It doesn't need to be killed off by WIFI for WIFI to be a success; and you don't need an IP infrastructure to move packets between your pockets. There's room for both technologies in the market, and it's the best solution for short range wire-free communication between devices that i've seen so far.
10:39:16 AM
|
|
On the OS Personality test, it turns out I am AmigaOS. "Ahead of your time. You keep lots of balls in the air. If only your parents had given your more opportunities to succeed."
10:38:59 AM
|
|
An interesting dimensional modelling issue: Does the Kimball-style dimensional modelling template allow the use of non star-schema structures within the data warehouse?
I worked once with a client who asked us to look at remodelling their data mart to make it more 'dimensional'. This is something we've done many a time and usually takes the form of analyzing the current data mart, picking out candidate facts and dimensions, pulling out hierarchies and putting it all together as a star-schema.
The only issue with this is that their reporting requirements were very complex, involved pulling data from all of our facts and dimensions in one go, using different sets of time ranges, and requiring lots and lots of aggregations. Doing all of this 'on-the-fly' with Discoverer, even using indexes and summaries, would lead to unacceptable query response times. In addition, one of their major requirements was that the data model is easy to understand and navigate, and doesn't require users to 'jump through hoops' to put together new reports.
It soon became clear that we had to create an additional set of data structures within the data mart that supported these complex reports, pre-aggregating the data over the required time periods and pulling all the required data from each of the facts and dimensions. But these tables weren't pure dimensions or facts, or direct summaries of them, so how did this fit in with Kimball's methodology?
I eventually found an article on Ralph's website entitled "Asymmetric Aggregation: What to do if your data mart isn't a shiny dimensional diamond" which addresses this issue. Although we'd already made the decision to go down this route, it was good to find confirmation elsewhere that this was a valid solution; the crux of the articles' argument (at least as it applies to our problem) is that not all aggregations are linear - they can become far more involved and time consuming because of complex business rules and this was exactly what was happening with the client.
I think the message from this exercise was not to get too hung-up on a perfectly dimensional model for the data mart, when an additional summary table, geared up to support a complex report that is 'mission critical' for the client, solves the problem easily. As long as the dimensional model is still the centre of the data mart, and the summary table is fed from this and used just for reporting, the original design ethos is still preserved but not at the expense of performance.
10:38:34 AM
|
|
Oracle have made a hostile bid for PeopleSoft. In the ERP market, PeopleSoft are the No.2, after SAP and Oracle, and just last week they announced an agreed takeover of J.D.Edwards, the No.4 in the market and particular specialists in the manufacturing market. It looks like the price Oracle are offering is too low and PeopleSoft's shares have already risen above that level after the bid was announced; the feeling is that Oracle will have to go up to $20-$25 per share to win the battle.
It's difficult to tell at this stage whether the PeopleSoft bid was planned or simply a defensive response to Peoplesoft's bid for J.D.Edwards, which would have given Peoplesoft the No.2 position and overtaken Oracle. What'll also be interesting is whether Oracle decide to continue with the J.D.Edwards acquisition. PeopleSoft are best known for their HR systems, and traditionally sell to large organisations, whilst J.D.Edwards usually appeals to smaller-to-medium companies. Certainly there was logic in PeopleSoft and J.D.Edwards linking up; what's not so clear is what Oracle would have to gain, apart from marketshare, and Oracle would almost certainly close down PeopleSoft and migrate their users to the e-Business Suite.
And what about Microsoft? Microsoft, although dominant in the desktop market, only have around 2% of the business software/ERP market, and that's after the buyout of Great Plains and Navision. Would Microsoft take this as the opportunity to buy the market leader, SAP? There's virtually no commonality between MS and SAP (if you've ever seen the SAP client interface, you'll know it's got nothing to do with MS, using it's own toolsets and widgets, with the underlying engine proprietary to SAP and data held in an abstraction layer that makes it effectively database and platform agnostic), but Microsoft could afford to buy SAP and it would give them a great platform to leverage Windows Server 2003 and SQL Server. Indeed, SAP recently made MS SQL Server their preferred database platform for the Windows platform, which made sense as every SAP sale that included an Oracle database was effectively making an additional sale for their biggest rival. About the only one who isn't interested is IBM, although they do have an agreement with J.D.Edwards and would miss the revenue.
This one should be interesting. Whatever happens now, PeopleSoft is effectively in play and they don't seem to keen on Oracle's advances, with history of bad blood and name calling between the two companies. PeopleSoft's own customers seem wary of the deal (but they would be, wouldn't they) and the ERP market, currently in the doldrums, is heading for a shake-up. There's an interesting discussion going on at Slashdot; More news to follow.
10:38:15 AM
|
|
An interesting couple of articles from Joel on Software on software development. He's basically given up on prototypes, figuring that if the prototype can do everything the product can do, it might as well be the product, and if it can't, it's not much use. The Iceberg Secret, Revealed talks about one thing every junior consultant needs to have injected into their head with a heavy duty 2500 RPM Drill: Customers Don't Know What They Want. Stop Expecting Customers to Know What They Want. It's just never going to happen. Get over it.
10:37:02 AM
|
|
A couple of excellent Japanese pet sites. CATPRIN, a tailor for cats. "Ever imagined dressing up your lovely cat into a fabulous beauty? You don't have to dress her everyday, in fact she might not feel comfortable with a dress on for days. Just dress her up only on special occasions like her birthday, takes a photo and that should leave you lots of memories and fantasies.". Includes a scary Chicken Costume Package and a High-School Girl Package. The Space Dog however, is completely unintelligable but it seems to be selling some sort of bizarre tinfoil spaceman hat for your dog. Superb.
10:36:43 AM
|
|
© Copyright 2003 Mark Rittman.
|
|
|
