IT Manager's Journal | Straight talk about building a Linux enterprise cluster with open source tools

Via Slashdot: IT Manager's Journal | Straight talk about building a Linux enterprise cluster with open source tools.

"However, Kopper experienced a steep learning serve in how to build a Linux enterprise cluster. Because there wasn't a book that could hold his hand throughout the process, he decided to write the definitive guidebook called "The Linux Enterprise Cluster" (No Starch Press), due out in January. Kopper recently spoke about the cluster he built based on Linux and other open source products."

The first SPIM lawsuit

Internet service providers file more spam lawsuits: "In the new round of lawsuits, AOL is suing numerous defendants and seeks damages as well as court orders forcing the alleged spam senders to give up their profits and cease their activities. One suit targets spam sent via instant messaging, also known as SPIM, the first such lawsuit, according to AOL."

Must be on ICQ (owned by AOL). Probably because it is so easy to send unknown users a message (since users are assigned and addressed by a sequential member number), it has been plaugued with unsolicited messages for years.

Optimal ETL Processing � Changed Data Capture Strategy | DMReview

Soumendra Mohanty writes in Optimal ETL Processing � Changed Data Capture Strategy about the specifics involved in capturing changed data in operational applications. Doing so comes in two flavors - reading log files for deltas, or capturing changes within the source system (and database) itself. Soumendra writes about the specific methods to utilize, and how Informatica and Oracle (9i and up) offer CDC as a feature. Useful info when dealing with very large source systems.

GMail Drive for Windows

I've written about this a couple of times in the past, and someone has figured out how to make GMail drives storage spaces for windows machines. Check it out:viksoe.dk - GMail Drive shell extension.

VMWare Oracle RAC Paper

Mark Rittman points to the new location of the famous Windows Laptop RAC paper, which has step by step instructions for installing a complete RAC environment using VMWare.

There's also a now-dated set of instructions for doing the same with a SQL Server active-active cluster.

DB2 Tuning Paper

Just came across this free paper on optimizing DB2 performance in complex environments. The author, Flavio Lima, is a consultant working at Bank of America and has some good tips.

EWeek: MySQL Update Puts a Toe Into Big Vendors' Database Turf

MySQL Update Puts a Toe Into Big Vendors' Database Turf has some info on the 4.1 release (available here) with some commentary:

At this point, MySQL is the world's most popular open-source database. In the spring, Forrester Research Inc. surveyed 140 large North American companies on their open-source plans and found that MySQL was high on respondents' project lists, with 52 percent of respondents reporting that they were using it or planned to use it.

Still, database giants such as Oracle Corp. sneer at the idea of MySQL competing. Open-source databases' share of the RDBMS market is negligible, with the following fears comprising the list of what scares CIOs away from the technology, according to research done by Garry:

• Informal product support.
  
• Rate of feature change, with some CIOs concerned that changes happen too slowly and others concerned about exactly the opposite.
  
• Lack of a product road map.
  
• Functional gaps.
  
• Licensing caveats.
  
• Low ISV support.
  

Many of these fears are irrational when applied to MySQL, Garry said. For example, the company has a formal support program and a product road map, and has secured a strong relationship with at least one ISV: SAP AG.

All that makes me think many CIOs need to choose who they get their info from a little better :)

What's in the 4.1 release, and what is coming up? The 4.1 release includes support of subqueries and derived tables. As for what to expect, from Marten Mickos, CEO of MySQL:

"Transaction capabilities have been in the production release for two years now," Mickos wrote in an e-mail exchange. The company is based in Uppsala, Sweden. "Stored procedures, triggers and views are in Version 5.0, which is out as alpha right now and will be in production release next year. Some of our customers have been writing stored procedures for 5.0 for several months already. "

Phil Windley's: DIDW 2004: Tony Scott on GM's Next Generation Outsourcing Model and its Affect on Identity

This is an interesting read, with the CTO of GM describing how they will deal in the future with contractor relationships, and some of the special considerations a company the size of GM has to address in its IT infrastructure.

From Windley's blog, where he is blogging the Digital ID World 2004.

Teradata Revamps DW for SOA, BI

Teradata Revamps DW for SOA, BI describes some of Teradata's upcoming features, including claims that it will make many third-party BI tools obsolete.

Lea adds, however, that Teradata doesn’t believe BI tools are going away—just that they’ll have a more precisely delineated niche: “I don’t think the BI space will go away by any means. There is still a need for reporting, pushing out reports, etc, and that will always be there.”

Third-party BI tools work just fine for these sorts of applications, Lea acknowledges, but it just makes more sense to perform them in the data warehouse, where everything—the near-real-time loading of data, event-specific triggers, analytics, and so forth—can be done with a single copy of the data. What’s more, he argues, it also helps business decision-makers—as well as leaders in specific business domains—get closer to the data.

Razza Solutions Announces 2005 Product Roadmap

PRESS RELEASE: Razza Solutions Announces 2005 Product Roadmap: "Innovated by Razza Solutions, Inc., the Razza Dimension Server (RazzaDS) is the world's only dedicated hierarchy management platform. The RazzaDS 2005 Product Roadmap outlines new features that will be implemented over the next year, including enhancements for Sarbanes-Oxley compliance, master data management, Business Intelligence application development and maintenance, and a new concept called matrixing that will greatly simplify the process of isolating specific, relevant attributes from very large data sets."

I've seen RazzaDS, a product to sync up like dimensions between data marts, pop up in a few places. AmSouth Bank for one. I have a feeling it will pop up again.

Mark Rittman: GIS-Enabling Your Oracle Data Warehouse

Wow Mark never fails to deliver the goods when it comes to explaining step by step how to use complicated concepts in the real world. GIS-Enabling Your Oracle Data Warehouse points to his new article about how to add geo analysis as a feature in a data warehouse, and how to use Oracle MapViewer in 10g to present the output.

As an FYI, taking advantage does require the EE option of Oracle Spatial, which runs about 10,000/CPU before discounts.

If you read this blog, you will probably find this interesting: InformationWeek > Data Demands Respect.

"We do three things really well," says Alex Dietz, who's referred to internally as "products and infrastructure technology leader" and functions as Acxiom's CIO (there are no traditional titles in the company). Those three things, Dietz says, are managing large volumes of data; cleaning, transforming, and enhancing that data; and distilling business intelligence from the data to drive smart decisions. That data is used by Acxiom's approximately 1,000 clients for everything from developing telemarketing lists and identifying prospects for credit-card offers to screening prospective employees and detecting fraudulent financial transactions.

...

No one at Acxiom seems to know exactly how much data the company manages in its 11 tightly guarded data centers. The company's central data center is located north of Little Rock, in Conway, Ark., where Acxiom was founded in 1969 as a spin-off of a local bus manufacturer. Acxiom also has data centers in Downer's Grove, Ill., outside of Chicago, and as far away as Sunderland in the United Kingdom. The company recently opened a data center in Phoenix, and an additional facility is under construction in West Little Rock. The best estimates are that the Conway systems alone store between 1.5 and 2.0 petabytes of data, or up to 2,000 terabytes.

Acxiom just passed a billion dollars in revenue for the first time last year.

Oracle 2020 — A Glimpse Into the Future of Database Management, Part 2 by Don Burleson

Oracle 2020 — A Glimpse Into the Future of Database Management, Part 2 by Don Burleson is Don Burleson's vision of the future of the state of databases and the responsibilities of an Oracle DBA in 2020. There's just way too much to extract out, so go visit and read all about it. Very interesting!

SQL Server Data Mining

SQL Server Data Mining is a new site put up by Microsoft's own SQL Server Data Mining team. There's some good information here, plus a few live demonstrations, including using Data Mining as a suggestion engine behind a site, and how to use Data Mining for data validation. Looks like they have done a thorough job.

Understanding Oracle OLAP Dimensions And Cubes

Understanding Oracle OLAP Dimensions And Cubes is a good introduction by Mark Rittman on how to access the Oracle OLAP features. Lots of code examples that ideally illustrate the how to implement the concepts of dimensions and cubes in an Oracle database.

Bookmark that page! High end Oracle features are considered a bit of a black art because its tough to find good examples, but heres one with it all laid out.

If you are familiar with Analysis Services, you'll notice quickly one of the reasons AS's market share continues to increase, while Oracle's OLAP share continues to decline. The knowledge level to effectively implement an OLAP soultion requires a skill level far, far above that required for Analysis Services (and without any difference in results.)

Microsoft spells out licensing approach for multicore servers

Wow... right on the heels of the Oracle position, Microsoft takes the opposite (and pro-customer) view.

Microsoft Corp. on Tuesday agreed to require only a single license for server software that runs on computers powered by a new generation of chips that squeeze multiple processors into a single package.

The processors in question aren't due out until next year, but by then I'm willing to bet Oracle will adopt the same position.

Oracle spells out licensing approach for multicore servers

Oracle spells out licensing approach for multicore servers:

"What is Oracle's position on software licensing with respect to dual-core processor systems?
We don't have a position with respect to dual-core processors. A core is equal to a CPU, and all cores are required to be licensed. Therefore, if you have a dual-core processor, you are required to have two processor licences. "

Just in case you were wondering.

There's also a dirty little secret in the Oracle world - never buy your licenses directly from Oracle, or an Oracle rep. If you do you'll pay list price, with standard discounts. The list Oracle discount schedule give you a 20% discount for a $250k invoice, and a 25% discount if you purchase over $1,000,000.

Go to a license releller to get the max discount up front (here's an example) with any size purchase, and pay less in upgrades and support, since that is based on purchase price. You'll get the same Oracle support, and still use the same Oracle contacts you always have, etc, just paying much much less. It makes no difference to Oracle who you buy your licenses from, and you get no advantage from purchasing them directly, at least not in my experience.

Microsoft pushes SQL Server 2005 back

Microsoft pushes SQL Server 2005 back: "Since early this year Microsoft said to expect the finished version of the product in the first half of 2005. Now the company has decided to release a new and 'incremental' beta called a Community Technical Preview (CTP) starting immediately, according to Microsoft senior vice-president Paul Flessner.
This means the finished version of the product will be pushed back to next summer."

Free MapPoint Developer Training from MSDN

Chandu Thota point out there is Free MapPoint Developer Training from MSDN, focused on using Mappoint web services. They are coming to a town near you in the US (hit the drop down list).

Databeacon makes a .NET OLAP Front end

Mark Levison posts First Commercial .NET No Touch Deployment application. Databeacon is my favorite Analysis Services front end, hands down. It is built to be web delivered, and the Java plug in is extremely intuitive and much easier to use than any other OLAP front end I've seen. Now, here's the .NET version, which looks very slick.

"No touch" is a bit misleading though. Instructions from his site:

1. Verify that Microsoft .NET 1.1 Framework is installed
2. Use the Microsoft .NET Framework wizard to grant full trust to trusted sites
3. In Microsoft Internet Explorer, add services.databeacon.com to the list of trusted sites (via internet options)

That wizard is under Administrative Tools on XP, under "Microsoft .NET Framework 1.1 Wizards", not Configuration. Then, the right option is "Adjust .NET Security", and on the "trusted Sites" tab raise the bar to full access.

And then under Internet Option in IE, add that site to trusted sites, but take off the checkbox first, or it won't work. What a pain... that the Java option is so much easier.

The front end looks very nice. I like how it shows both the report view, and whatever graph view you choose at the same time. This is a real improvement. Here's a screenshot from the example:

Free copy of Visual Studio .NET

This happened before, it looks like it is up again - ASP.NET Webcasts Series has details on getting a free copy of VS.NET (for VB.NET) plus an ASP.NET book, plus a couple of extras CDs, all for watching a few free webcasts.

It's too good to pass up. Especially if you want to experiment with Reporting Services, you need a copy of VS.NET, this is a perfect way to pick it up for free.

Configuring SQL Server Express 2005

SQL Express 2005, a beta of the next version of the MSDE database, is free for download from MS and interesting for all of the Yukon features that have been included. It has been announced here and several other places in the past.

I never gave it a test drive myself, but I have a friend who is using it for development, and it turns out it is a little short on documentation, specifically on how to connect to and administer the server. Configuring SQL Server Express 2005 is a great guide to doing so. To frame the pain:

turned on my home computer and downloaded the package of Web Dev Express, ASP.Net 2.0 Framework, and SQL Server Express. I installed all three. My excitement still ran strong while I fired up Web Dev Express for the first time.

Every trace of my excitement was smashed and crushed as I clicked on Database Explorer, tried to connect to SQL, and received my first (of many) error message. Within a relatively short period of time, there was no excitement left, simply frustration, but thankfully a good amount of determination.

The following is the result of my entire weekend spent agonizing, searching, yelling, troubleshooting, screaming, and finally exulting in my efforts to configure SQL Server Express 2005.

Seems to me this could have been handled much better by Microsoft.

Native Web Services in SQL 2005

Native Web Services in SQL 2005 on Ajarn's SQL Corner talks briefly about native web services support in SQL 2005.

The capabilities in one form or another have been there for awhile in SQL, and in Oracle, to connect via some sort of web service, but it never has really caught on. There is a performance penalty, and I imagine that is probably the reason. What would be really interesting is if everyone agreed on a single standard to connect via a web service, then this might get some traction.

Sun and Sybase claim data warehouse record - vnunet.com

Sun and Sybase claim data warehouse record - vnunet.com: "With a trillion rows of data, Sun and Sybase claim that the warehouse can hold enough data to track the history of all world financial trades on all stock exchanges, or hold enough data to track all credit and debit card transactions in the entire world for the past seven years."

Get Real: HitMaps comes out of the closet

Get Real: HitMaps comes out of the closet points out the HitMaps project, which collects all the IP addresses from your web site visitors, converts them to geographic locations, and plots them on an image of a world map. Any website can post it, as I have on the right column. Cool stuff!

And not all that technically challenging with the right tools. Using the same principles, you can apply this to call center stats, customer records, or anything else under the sun.

Steve Lombardi - Driving around Mapoint Location Server

Steve Lombardi - Driving around Mapoint Location Server is a 20 minute video on Microsoft's Channel 9 site that is an explanation and demo of Mappoint Location Server. Mappoint Location Server is very cool - even though it has the Mappoint name, it has little in common with Mappoint 2004. The purpose of the server is to track the location of cell phones, presumably for corporate employees. From there they can get plotted on a map, and you can do stuff like find who is the nearest to a particular location.

There's an explanation that the accuracy depends on the carrier. Some only provide the location of the nearest tower, while others use built-in GPS. Currently, only a few cell phone networks are supported (only Sprint in the US, BellMobility in Canada, and O2(?) in the UK).

Analytical SQL Functions in Oracle

Looks like a good start to a series of articles about using Oracle's analytic extensions. � Analytical SQL Functions - theory and examples - Part 2 on the Order By and Windowing Clauses follows up on Pat 1, which discussed PARTITION.