Introducing Yellowfin Business Intelligence
The focus of my company up to now has been consulting with Cognos Business Intelligence software. However, I am looking to expand my entrepreneurial horizons, and to that end, I am currently working on a partnership with Yellowfin, an upstart BI software developer based in Melbourne, Australia. Their flagship product, Yellowfin BI, emphasizes usability and making business intelligence easy: easy to configure, easy to administer and most of all, easy to use. (More about their design philosophy can be read here.) I believe that the time has come for this type of product in the BI market for the following reasons:
- Compared to most enterprise software packages, the UI is sublime.
- 100% web based, with a zero footprint web browser client.
- It is built on open source technology with open standards.
- No hand coding of SQL is required.
- A rapid deployment is not only possible, but it can be expected.
In short, Yellowfin is simple, secure and scalable. All of this adds up to a faster path to business value.
Yellowfin is simple, secure and scalable.
What I see is the greatest advantage of Yellowfin is the shift from the traditional BI project thinking. Traditionally, this entails an attempt to build an all encompassing data warehouse, followed by countless hours of meta-data and report development. The pitfalls are plenty in this approach. Yellowfin BI is not a panacea, but the software lends itself to a more agile approach to BI development. While it is robust enough for an enterprise deployment, it is ideal for the SMB or an isolated division of a larger organization that is not well served by the central IT department.
Yellowfin has been successful in the Asia-Pacific region but the company is only beginning to make a splash (pun intended) in North America. It has been selling well as a front end for Sybase IQ, and Yellowfin is currently expanding the partnership with Sybase globally. Historically, the Sybase BI partner of choice has been Microstrategy, but I expect that it is only a matter of time before Yellowfin gains the upper hand here.
If you interested in learning more about Yellowfin, please contact me, especially if you are in Canada and you are searching for a simple, secure and scalable BI solution.
Using SQL Server Express as the Cognos 8 Content Store
Microsoft SQL Server is fully supported by Cognos 8 Business Intelligence Server, but the Express Edition is not officially supported for use as the Cognos 8 content store, as per the support website. If you try anyway with a default install of SQL Server Express, you will see an error like the following in the cogserver.log file when the Cognos service is started:
192.168.0.103:9300 5496 2008-06-19 15:16:24.390 -5 Initialization_SESS Initialization_REQ Thread-16 CM 888 1 Audit.cms.CM StartService Warning CM-CFG-5063 A Content Manager configuration error was detected while connecting to the content store. CM-SYS-5003 Content Manager is unable to access the content store. Verify your database connection parameters, and then contact your database administrator for assistance. Failed Logon:com.jnetdirect.jsql.JSQLException: TCP/IP connection failed to host:localhost java.net.Con nectException: Connection refused: connect url:jdbc:JSQLConnect://localhost:1433/cognos83
This is a generic error that states that the Cognos 8 application cannot connect to the content store database. However, with a few configuration tweaks, this can be resolved. By default, the SQL Server Express networking protocols are disabled, which means that it will not accept TCP/IP connections. This is why Cognos cannot make the connection and throws the error. To resolve the problem, we need to do two things with SQL Server Express: first, enable TCP/IP, and then second, specify a port number to listen on. Here are the steps:
- Launch the SQL Server Configuration Manager.
- In the left hand window tree view, expand SQL Server 2005 Network Configuration.
- Select “Protocols for SQLEXPRESS”.
- In the right hand window, right click on “TCP/IP” in the list of protocols and choose “Enable”.
- Right click on TCP/IP again and select properties.
- Click on the tab labelled “IP Addresses”
- Scroll down to the bottom to view the section titled “IPALL”
- Clear the value for “TCP Dynamic Ports” so that it is blank.
- Enter TcpPort number to use when making remote connections, for example, “1433″.
Restart the SQL Server service. Now, when you start Cognos 8, the connection will be made and the content store tables will be created in the database. You can verify this by watching for xml files being created in the logs directory (with names like 001.CMCreateWSRPObjects_results.xml), or by refreshing the tables folder in the SQL Server. I saw 116 tables created, but I know that can vary, according to the version of Cognos 8.
Caveat emptor: Keep in mind that SQL Server Express is not supported by IBM Cognos for use as the content store database.
Thanks to the SQL Server Express Weblog for the pointers that helped me get this working.
How to Get Started with Operational Business Intelligence
Getting started is often the hardest part of any project because it involves dealing with so many unknown factors. Operational BI can be especially daunting, as it can have such an impact on non-technical people and processes. Claudia Imhoff, on the b-eye network blog, offers some thoughts on getting started with operational BI:
1. Start small - Making changes, especially changes that impact people and their business processes, can be tough. By starting small and building on subsequent successes, everyone gains confidence in the new system, paving the way for bigger changes down the road.
2. Asses the existing DW/BI infrastructure - As BI moves into an operational role, existing problems will only get worse, and they gain a wider, less forgiving, audience. Make a plan to reduce bottlenecks and solve any delivery, infrastructure and data quality issues.
As BI projects go, introducing BI software into an operational role is particularly challenging. However, the payoffs can be huge for both the organization and the IT department. The organization wins with gains in productivity and efficiency in achieving business goals, while the IT department wins increased respect as a strategic part of the organization.
Business Intelligence gets Cloudy
Cloud computing is a great IT buzzword for this year, and sure enough, there are some BI vendors hitching themselves to this bandwagon. However, beyond the hype, there are some real benefits to hosting BI applications outside the organization’s own data center and “in the cloud”, as explained in this InfoWorld article:
- The client interface is web based - End users are already familiar with the web browser, and increasingly, they are at ease with using web applications.
- The IT department’s role is simplified - There is no additional server infrastructure to maintain, and there are no desktop applications to roll out, troubleshoot and keep updated.
While there are other issues to address, I think the weight of these two arguments, which both reduce the overall cost of BI implementations, will continue to drive more BI software into a cloud computing model.
Are Analytics the future of BI?
While analytics have been around for a long time, most BI implementations focus on building the data warehouse and delivering the data in a nice format. For analysis, OLAP cubes are created, but often, the serious analysis work is done by exporting the data to Excel. The major BI vendors have tried to branch out into analytics, offering packages, usually for an ERP system, built with their own BI tools. The exception is SAS, which has focused on analytics, and pretty much owns the higher end of the market.
However, there are some upstart BI vendors attacking the weak analytics offerings in the existing BI solutions. NextAnalytics is one such company, founded by Ward Yaternick, a former Cognos employee.
In a recent blog post, he sums up the problem his software solves: “analytics are iterative and sequential.” This makes analytics very expensive to do and to maintain with the current BI tools. NextAnalytics also has a non-traditional licensing and pricing model, similar to other commercial open source products, which will be a breath of fresh air for those used to negotiating with the big BI software vendors.
7 Steps for Troubleshooting BI Software
Earlier today I published the final version of this white paper: 7 Steps for Troubleshooting BI Software. Have a look, and if you find it helpful, please say thanks by sending me a note or by leaving a comment.
Five Security Questions to ask about your BI Project
Security is often an afterthought on BI projects. The conventional thinking, as encouraged by BI vendors, is that the BI implementation will leverage the existing security infrastructure, be it LDAP, Active Directory Server, or otherwise. While that might be the case, thinking about security cannot end there. The following five questions need to asked, and answered, early on in any BI project:
- How will penetration testing be done?
- How are user IDs and passwords managed? (Single sign-on is the preferred answer.)
- What is the encryption policy? 128-bit SSL encryption is typical, but sensitive data might require more.
- Are the servers shared with other applications? Are the servers virtual servers? If so, what is in place to prevent data leaks, malicious or accidental?
- What is the backup and disaster recovery plan?
As BI moves outside the firewall, onto mobile devices, and perhaps running on servers hosted by a 3rd party, the questions around security need to be asked early, and if necessary, often.
Nine Steps for an Effective Change Managment Process
Once a business intelligence system goes “live” and into production, a change management process is an essential tool for keeping production servers humming, the jobs running and the business users happy. A balance must be struck between being strict enough to maintain control and accountability, but agile enough that changes can be made quickly for the business. Here are the nine steps for an effective change management process:
- Change Request
- Change Approval
- Planning - includes the evaluation of any risks and a back-out plan
- Testing - simulation in a lab or on a non-production system
- Scheduling - can the change fit into the regular maintenance window?
- Communication - all affected groups need to be notified
- Implementation - actually making the change
- Documentation
- Follow Up - closing the loop with all involved parties
As business intelligence and data warehouse applications move into more mission-critical, operational roles in the enterprise, change management is an essential consideration.
Tool #1 in the BI developer’s toolbox
If you are an IT professional working with business intelligence software, or any software for that matter, than you accumulate a collection a of software tools that help you get the job done. Tool #1 is the text editor. Text is software’s raw material, used when coding, scripting, configuring and documenting. The default text editor for Windows is Notepad, and it is what most people end up using for everything. However, there are many better options. The importance of a text editor in the IT industry is evidenced by the massive collection of open source editors available on SourceForge and Freshmeat.
The Pragmatic Programmer offers some guidelines for making an informed choice of text editor:
- Configurable - the look and feel must be comfortable, and keyboard shortcuts must be available
- Extensible - integrates with other tools (like Cygwin or Perl), has a plug-in architecture
- Programmable - ability to automate multi-step tasks that have to be repeated ad infinitum
- Syntax highlighting
- Auto-completion
- Auto-indentation
- Templates
- Access to help systems from within the editor
- IDE features - compile, debug, preview
The best advice the authors offer is simply this: “Use a Single Editor Well” (Tip 22)
My current favorite editors for Windows are Metapad (which replaces Notepad) and Notepad++. For Unix/Linux, I like VIM. I rarely work on a Mac, but if I did more often, I would surely use TextMate.
10 steps to ETL
Whether you buy or build your ETL tool, ETL development can be broken down into 10 steps:
- Create a high-level design for the target model.
- Choose, install, configure and learn the ETL tool of choice.
- Determine your default strategies.
- Drill down by target table.
- Develop the historical load for dimensions.
- Populate historical fact tables.
- Develop the dimension table incremental processing.
- Develop the fact table incremental processing
- Implement aggregate and OLAP loads.
- Create the plan for ETL operation (ie, production support), automation and future enhancements.
Thank you for the great post Tod McKenna.