Comparing new DB offerings from MS, Oracle, and IBM

Just stumbled across Keep an Open Eye � BI is marching to Microsoft�s Giveaway Drum, where the author gives his first impressions of Oracle 10g, DB2 Express, and SQL Server.

All do their jobs pretty well, and 10g and DB2 include some BI freebies that were not there before. Looks like an interesting blog.

Comparing Informatica And OWB

Mark Rittman has an extensive writeup on how Informatica compares to Oracle Warehouse Builder, Oracle's included ETL tool. The biggest difference is that OWB only supports Oracle as a destination, but the article goes into much more detail and is a great read.

SQL Server's DTS is the step child in the ETL tool field, although it has much more functionality than many people realize and is ideal for many tasks (I've seen it process millions of records for a Siebel system reliably on a nightly basis). It is not an ideal enterprise tool however, as it lacks some things expected on such tools, like Metadata integration, compliance with OMG standards, and the fact that many essential features are turned off by default, and must be activated on a per-package basis (such as logging and error alerting). It does have some advantages however, mainly easy bulk loading of flat files, custom tranformations, a built in object to interact with analysis services cubes, and the ability to load anything accessible with an OLE or ODBC driver, from any available database to Excel spreadsheets to text files.

One of the top new features in SQL Server 2005 is supposed to be the beefed-up DTS, which the developers claim will compete with any commercial ETL tool. Considering the cost of many ETL tools, it will be something to look at very closely.

DTS has usually done the job in projects I've worked on, but that's probably because very few companies have tried to implement any sort of metadata management across the enterprise. In such a case, Informatica definitely fills a need few others can match presently.

If you aren't familiar with Metadata and how it fits in the big picture (and very very few people do), or how the OMG is trying to standardize it, take a look at this presentation.

Building Advanced Reporting Services Applications

Reporting Services is hot in the SQL Server world - it's proving to be a reliable replacement for more expensive reporting suites. Building Advanced Reporting Services Applications has some advanced tips.

Also, Service Pack 1 for RS has been out for about a month.

World's biggest computing experiment

InfoWorld has information on the CERN Paticle Accelerator, currently scheduled to go live in 2007. Oracle is a participant, but the article focuses on IBM and the failure of its product to handle the enourmous storage demands.

One of the Grid's biggest hurdles is however what to do with the huge amount of data produced by the particle accelerator. It is calculated to pump out 15PB of data a year (or 34TB a day) continuously. Yet at that rate, IBM's current storage solution would be able to store just 20 hours of data. The situation will get worse over time as well, with the IT head of the project, Wolfgang von Ruden, estimating that by 2010, 100PB a year of data will come from the LHC project.

Wow that is a ton of data... the system will need to have the capability to analyze 40 million particle collisions per second, 24 hours a day, for at least 13 years.

InformationWeek > Open Source, Open Questions

InformationWeek > MySQL > Open Source, Open Questions is a recent article in Information Week profiling several companies, including Sabre and PriceGrabber.com, that have made the jump to MySQL.

An interesting spin in the article is:

Some companies find they outgrow MySQL. Fairfield Language Technologies, which develops Rosetta Stone foreign-language learning software, used MySQL to provide content to its Rosettastone.com and online language-learning centers, as well as to manage licenses of online customers. But the company recently decided it needed the scalability and server redundancy of Oracle's database-clustering technology. Fairfield will still use MySQL for several in-house applications, says Robert Bland, VP of operations, finance, and IT.

A sidebar to the story talks about CA's release of Ingres to the open source community. According to the article, they did it to "catch up" after a decade of neglect of the code base.

Using IM for Data Transport

Feedster delivered me Burnham's Beat: DIM: Hijacking IM for Data Transport which has a pretty concept: using IM for data transport. Despite the enthusiasm of the author (who happens to be a VC investor in the company he's pimping for) I see some serious drawbacks to this on public networks. First, there's a message throttle, that only x messages can be sent in a certain timeframe, and second, instant replies are a sure way to get tagged as a bot on the network and get kicked off for 10 minutes or so.  About the only advantage it has is that one of the "agents" or sources is accessible from anywhere, and can connect to the internet from anywhere, since name resolution is handled by the IM network itself. Still an interesting idea, although I think web services have all the advantages and none of the disadvantages, at least when it comes to the business world.

A few years ago every DB vendor moved quickly to make SOAP interfaces to their offerings, it would be interesting to see a generic database connector for an IM network, although I doubt it would be very practical. I created  mini-IM-database-admin style app about a year ago as a proof of concept style project, but it was meant for humans to control, not an application.

Another intro to Yukon Data Mining

One of the Yukon developers announced on the Analysis Services:Data Mining MSN group that a recent presentation on Yukon Data Mining has been uploaded to the MS site. Not much new information if you've seen one of these before, just reinforces that MS is targeting the Data Mining industry with the next release of Yukon.

Enterprise Performance Management: An Executive-Level Implementation Guide

How about a BPM/CPM/EPM (let's just call it xPM) morning?

Via DMReview, Enterprise Performance Management: An Executive-Level Implementation Guide has a nice plan to follow with all of the things to think about when building a performance management system for a large company.

Also, in a bit of a gift for developers wishing to get a handle on BPEL and how all the pieces fit together, ActiveBPEL has released their BPEL engine (written in java) to the open source community. It claims to fully implement all of the BPEL spec.

BPM Overview on InfoWorld

InfoWorld: The click-and-drag enterprise is a very good overview of BPM, and how BPEL fits in to the picture. Here's an excerpt:

“A business process needs to be completely defined independently from IT,” says Matt Quinn, director of technology marketing at Tibco, which acquired BPM pure-play Staffware last month. “It needs to be a business-focused view of what a business process constitutes. Of course, you can’t simply drive everything top-down — you do need to actually capture business processes at a technology level as well. But if you understand the business process from end to end, without having the constraint of IT initially, you’ve got a much broader context for really understanding what your business is.”

Going Deep on Business Activity Monitoring with BizTalk Server 2004

Going Deep on Business Activity Monitoring with BizTalk Server 2004 is a MS TechEd presentation. The data on Biztalk is a bit sketchy, but the explanation of BAM and the details of how to implement a BAM solution are spot on.

Defining Alphabet Soup

eBPML.org has an informative article up titled Defining Coordination, Composition, Choreography, Collaboration, Orchestration, Transaction...
. It attempts to explain what all these terms mean in the Web Services stack. I found it very useful. When I used to think of web services, I thought of those magnetic poetry sets that have 50 words you can move around on your refrigerator and make them say meaningless yet interesting things. However, after this article I am now fully convinced that there is truly a need for all of them.

RFID wave coming

Tag You're It from Technology Review has an article about some creative uses for RFIDs, including the recently announced Delta Airlines plan to tag all luggage with RFIDs, and the example of Coors UK, which bought into a plan to tag all their beer kegs used for delivering beer:

"RFID implementation by the British breweries also had some positive unintended consequences. Because the technology provided an audit trail for each keg, the breweries were able to claim tax credits on the amount of beer left in each keg. Typically, a brewery is taxed on the amount of beer shipped out. With an airtight audit trail now in place, breweries weigh the kegs upon their return and receive tax credits on the bottom swill or, if the keg was defective, the full keg. 'Companies save roughly $1 to $12 per keg, depending on how much beer is left in the container,' says Thomas Ryan, an analyst with the Aberdeen Group.

The tax credit alone paid for the early implementations of this system,� says Adams. "

Interesting article. The mountain of data produced by RFIDs and their audit trails will have mountains of knowledge for cost savings, process reengineering, and possiblyy even revenue generation.

Upgrading to Oracle10g

Lessons Learned: Upgrading to Oracle10g has anwers and perspective on upgrading to Oracle 10g. Links to many other artciles by Oralce gurus with additional information.

A little off topic, and noting the copy of eWeek in front of me that has a huge ad on the back cover ad that says Oracle 10g runs on 64 node PC clusters, I would have liked to have asked the question - does one of those actually exist? It's all in good humor, seriously, it's just that I've never met an Oracle sales rep that knows of any customer that is even considering it. A couple of nodes, on extremely expensive hardware, sure, but the promise of a commodity architecture isn't quite there yet. Stuff like that always makes me a little wary of Oracle-the-corporation's messages.

JOLAP, or Java's attempt at an open way to query OLAP

OLAP - JOLAP has one developer's experience with JOLAP, which is all too familiar. Mondrian is an open source OLAP engine (the only one I know about), and is about the only engine popularaly known for supporting the JOLAP spec (although there are several commercial engines that do, including Oracle and Hyperion, among others). However, even though it is what will pop up in most search engines, it doesn't really support JOLAP yet, grudgingly leaning toward MDX and XMLA.

Mark Rittman in the pas has written up some good analysis on JOLAP and XMLA.

JPivot is the preferred (and open source) front end for Mondrian, and it too supports MDX fairly well.

I wish there were more open source alternatives for OLAP front ends (not to mention back ends). It seems to me an engine to create ROLAP friendly output is attainable by a grassroots effort with available tools, but no one has decided to bite that off yet.

Building an IM client with C#

This is a cool webcast for someone so inclined:

Less than 12 hours to the RTC Building Webcast...
A full chat client UI will be developed usign the controls right in the designer. You'll get to see the application being created right in front of your eyes. The backup project with the finalized code is still there, but hopefully it won't be needed right ;-)
With the client out of the way, see how various types of features can be added into the protocol layer to enable enhanced functionality that only Basic Messenger users will be able to take advantage of.

Financial Services and Oracle

Banking on Business Intelligence: Financial Services Leaders Do More, Sell More with Oracle is a pdf with about a dozen articles about how Banks are using Oracle applications in strategic roles. Very professionally done, but pretty much a marketing brochure with no technical content.

The article on BI focuses on how GE Capital uses Oracle Financial Analyzer. They like it.

MapPoint 2002/2004 OLAP Wizard Add-in

A new version of the MapPoint 2002/2004 OLAP Wizard Add-in has been posted on Microsoft's download site.

Managing a Music Collection Using Visual Basic Express and SQL Server Express

For a crash course on SQL Server and .NET development with an easy yet interesting project, complete with code, check out Managing a Music Collection Using Visual Basic Express and SQL Server Express, a new article on MSDN. Shows step by step how to create a little mp3 player and catalog app from scratch. And you just have to admore someone that uses an Iron Maiden album as their example.

Hyperion Business Intelligence Review

Over at BPM today - a Product Review for Hyperion Business Intelligence. Brief summary - It does everything you can dream of and has no weaknesses. If you have to ask the price, you probably can't afford it.

Excellent introduction to Reporting Services

Have you used reporting services yet? This free add on to SQL Server 2000 is all the rage for alot of DBA shops lately.

If you still haven't got your feet, I just stumbled across this archived web event from Microsoft. Covers installing, configuring, authoring reports, managing reports and delivering them over the web, email or other tools. Step by step with all screens shown, very good walkthrough.

Banks, Brokerages Dogged by Message Storage Rules - Computerworld

From the IM world: Banks, Brokerages Dogged by Message Storage Rules - Computerworld. Very real problem with the challenges of archiving messages of new channels (like IM) in a speedy manner.

There's several ways to approach this. Companies that have a private, onsite IM network (such as Oracle IM, Facetime, and many others) have logging features built in. Oracle's solution stores messages in an Oracle database, obviously.

Communication on public networks is a little different - the best way to handle them is to have a gateway such as IMLogic or Akonix and log everything coming in or out. And seriously, every company has employees that use MSN or AIM or Yahoo IM. There are many, many clients that have built in, cofigurable features to tunnel through HTTP or SOCKS proxies. If employees can get to the web, they can get to an IM network.

SQL Server Express, ie, a public Yukon preview

Well not a full one, but there's alot of talk the past few days about the new beta of SQL Express (formerly called MSDE.) SQL Server Express has alot of the Yukon features built in, including full support of the .NET CLR.

For those unfamiliar, MSDE (now called SQL Express) is a free, watered down version of SQL Server that many, many applications install with their product and use as a data store. Except for the built in restrictions of memory and CPU usage, it is pretty full featured, and can be managed from and SQL Server admin tool.

Want to know just how many applications install MSDE? Download these Microsoft Security Tools (designed to spot SQL Server installations vulnerable to the Slammer worm, and patch them) and run sqlscan.exe on your network. Odds are you'll find many more MSDE installations on desktops (many unpatched) than actual SQL Servers in your company.