Blog: William McKnight

« October 2005 | Main | December 2005 »

November 30, 2005

Bed time for business intelligence professionals

According to "Missed ZZZ's, More Disease?" in Science News, skimping on sleep may be bad for your health.

While it's too soon to say moderate sleep deprivation can damage your health, definite hormonal changes occur when people get less than 8 hours of sleep. Business intelligence professionals are probably more guilty of this than most.

Research indicates sleep deprivation can increase risk of obseity, cancer, diabetes and heart trouble.

I think I'll quit blogging and go to bed.

Share:

November 28, 2005

Today's News: INFA #1 in Gartner Market Report

Gartner has named Informatica the leader in the data integration market, securing the number one position for the fourth straight year in Gartner’s ETL market share report. Informatica captured 25% of the ETL market in 2004, with an increase in revenue of 10% from 2003 and over a 5% increase over the nearest competition, Ascential/IBM.

Informatica is ripe for acquisition and its purchase will (would?) send shockwaves through the industry. It won't be IBM since they now have Ascential. If Microsoft, it would signal a real enterprise push for them and I would expect more acquisitions from them soon following. It won't be Business Objects since they are strongly pushing their Data Integrator (formerly Acta) product now, but it would be interesting to see if a "BI company" could do this (versus a big player like IBM, Oracle, Microsoft, SAP). SAP has rolled out Netweaver so I'm discounting that possibility.

PeopleSoft EPM, the out of the box BI suite from PeopleSoft, ships with Ascential. This may change. The Oracle-Informatica partnership around that could lead to something more.

And then there's Microstrategy. Informatica and Microstrategy could power an entity from even humble beginnings right into this top-tier.

Of course, these are my opinions only.

Share:

November 21, 2005

The end of the CIO?

The premise of this article in E-Commerce Times is that the CFO is so responsible for organizational data (i.e., compliance) that, in some shops, s/he might as well run all of IT. Wow! I'm not sure where this "consensus" referenced is coming from, but I have not encountered this. We've come so far in terms of legitimizing the CIO position away from being pure support of any department and/or reporting to the CFO. Most CIOs are peers of CFOs.

Despite the obvious importance of Sarbanes-Oxley, I don't believe this is a trend. CIOs must be sensitive now to not only internal knowledge worker clients supporting customer needs, but to internal compliance requirements. This is a natural step for most CIOs. The position has developed strong business acumen over the years and should be able to work closely with the CFO to deliver on compliance requirements without the entire function going under the CFO.

Share:

November 18, 2005

Live to a ripe old age - with a little help from your friends

I found this study interesting since we are getting close to that big family holiday, Thanksgiving.

Want to live a long time? Make some friends.

FRIENDS, not family, are one of the keys to a long life, a study of elderly Australians suggests as reported in New Scientist.

The research suggests that friends offer advice without pressure, which might help people take care of their health without adding stress.

Share:

Gartner Clients Consider BI to Be a Major Priority

Garner came out with a research paper today titled "Clients Consider Business Intelligence to Be a Major Priority" that states "business intelligence jumped from the No. 10 slot to the second-highest priority on CIOs' agendas."

Some observations:

· Small-medium businesses (SMBs) make up a large growth market for BI software

· In the key verticals of manufacturing and financial services, BI accounts for the highest share of software budgets — 18 percent and 16 percent, respectively.

· The education market is a growth market for BI - projecting that BI will make up approximately 10 percent of the total software budget, mainly as a result of recent laws and regulations involving education, such as the "No Child Left Behind" policy in the United States.

· Other markets that are just beginning to adopt BI on a larger scale include healthcare and government.

Drivers of BI growth:

· To respond to user needs for timely data

· Efforts to improve decision making

· As a response to governmental regulations.

· The continuing need to reduce total cost of ownership within organizations

Share:

November 15, 2005

Focus: Disaster Recovery

I remember the day, not many years ago, when data warehouse programs would annually undergo a disaster recovery plan test. This usually involved a weekend and usually involved some important revelations in terms of readiness! I've noticed in the past year that those tests are seldom done anymore. Not being a server expert, I assumed that there were some important new built-in capabilities to servers or more failover environments in place that removed the need for the exercise.

So, I began to make it a point to find out more and, while surely the servers have improved in this area and there are more failover arrangements in place, the primary reason data warehouse disaster recovery plans aren't done as much seems to be that the exercise is out-prioritized. This is not a great reason. Of course, averting a disaster is always a tough justification in light of the many clearly progressive things we like to do for our data warehouse environments.

But consider these simple things that form the basis of a sound disaster recovery plan, as given to me by a systems expert...

1. Have a backup strategy that you execute; usually this will mean daily backups of all production servers (my expert discouraged incremental backups) and hourly log backups
2. Diagram the network in detail with specifications and contact numbers (incidentally, CSI has capabilities here we call ClarityPath)
3. Prioritize systems and connections by acceptable downtime; this prioritizes the recovery effort focus
4. Build a recovery team including these roles - project leader, communication leader and technology experts who are on call 24 x 7
5. Test the plan (the premise of this blog entry) which includes testing those old tape drives to make sure they can read the old tapes still needed
6. Review the plan systemically once per month - make it a living document
7. Put up a secure website where status can be communicated in the event in-house systems are down and people are working from off-network computers

It's obviously difficult to evaluate these team members on the basis of something that will likely and hopefully never occur. Therefore, readiness metrics are more appropriate for the evaluation.

Share:

November 13, 2005

Source System impact on DW/BI data model

Today, I'm thinking about building the DW/BI data model and the various approaches builders will take on the relationship to the source system(s). This is one of about 35 different aspects of a DW/BI methodology that are often effected in a shop in de-facto fashion, and often suboptimally. With this entry, I'd like to point out the importance, not only of this decision, but of making all these decisions in a heads-up fashion.

Theoretically, the source system(s) should have little impact on the DW/BI data model. However, let’s take a look at real world data warehouse model building strategies.

Data warehouse data models are built with one of 3 strategies in mind...

1. Mimic the operational data model(s) completely. This saves time in the modeling process and you know that you are building something for which you will have the data for and you know where you’ll get the data. With this approach, you solve one problem you are probably having and that has to do with limited query access to operational systems.
2. Model the data warehouse based entirely on user access requirements irrespective of the source systems and the availability of data. This will create a data model that is fit-for-purpose and possibly technically elegant, but the data acquisition piece will need to then be tied in with the modeling effort. Since the source data has not been considered, the mapping effort may be long and there may be unrealistic constructs in the model.
3. Model the data warehouse based on user access requirements but with knowledge of and an eye on the source systems. Data acquisition design is a parallel activity in this strategy with source systems identification and analysis having previously been done so the modeler understands what is available and doesn’t stray beyond the bounds of what is available during the modeling effort.

I recommend the third approach for DW/BI modeling efforts. This strategy strikes the best balance between designing for user access and building that which you can reasonably populate. This illustrates the need for a close working relationship between the data modeler(s) and the data acquisition programmer(s), whose work eventually needs to be brought together.

Share:

November 11, 2005

Management Phobia in Business Intelligence

From interviews with hundreds of BI professionals over the past year as part of my environment assessments, there are a few conclusions I can draw. One of them is that, as compared to previous years, more people don't want to be a manager and this includes managers. I suggest a variety of factors leading to this:

1. Managers have too many meetings

Self-explanatory

2. The people who work for you have higher expectations of what you need to do for them

You get to play counselor, psychologist, motivational speaker and, most definitely, personal advisor. These aspects of the job can be quite a change from the technical work.

3. Recent technology innovations provide an attractive career path on the technical side

Real-time data warehousing, EII/EAI, ERP integration, master data management, managing multi-terabytes of data, diverse and growing user communities, etc. all keep the BI technical professional out of persistent doldrums.

4. Technology changes make it seem difficult that you can get back on the technology track if you wanted to someday.

This has resulted in very careful consideration of the management track on the part of many BI professionals.

5. Long hours as in being told 55 hours per week are not enough

Self-explanatory

Management has always had its difficult challenges, but in the past those were acceptable based on the financial and career upside. However, in today's economy, with restructuring activity rampant, many of the hierarchical structures of the past, and the perceived benefits of moving into management, have been eliminated. With merger and acquisiton activity, managers now also supervise across state and country boundaries and across functions, exacerbating the stress (and lowering job security).

If you have any opinions as to whether, in your experience, this is true, please post.

Also if you have an opinion as to whether this is hurting or helping business intelligence programs, post those here as well.

Share:

November 9, 2005

Upcoming webinar on Master Data Management

On November 17th at 2:00 ET, I will be giving a webinar on the B-EYE-Network with Mark Smith of Ventana Research titled "Justifying and Implementing Master Data Management for the Enterprise." You can sign up for it at this link.

Here is the abstract:
More complex and demanding business environments are forcing more heterogeneous systems environments. Data integration is required to synchronize master data to get a single, consistent view of an enterprise’s core business entities like customers, products, suppliers and employees. Effective master data for any subject area requires input from multiple applications and business units; and enterprise master data needs a formal management system. Business approval, business process change and capture of master data at optimal, early points in the data lifecycle are essential to achieving true enterprise master data.

-An introduction to Master Data Management, which enables data synchronization across the heterogeneous system environment independent of operational and analytic domain

-How MDM directly impacts the business through facilitating more accurate and timely business decision-making and improving application delivery ability, all at a lower cost

-How to organize the enterprise, both business and technical roles, for the determination and implementation of master data rules

-How to take the conformed dimension concept to earlier points in the data lifecycle through MDM Architecture choices

Share:

A Free DBMS? From Oracle?

That's right. This press release uses "free" and "Oracle" in the same sentence. Perhaps feeling the heat from SQL Server 2005, Oracle has plans to release a free version of its database by year end to compete in the SMB market, suddenly becoming a concern in the BI marketplace.

Obviously, there are going to be limitations. The database is limited to 4 GB of data, 1 GB RAM and 1 processor. The goal is for this "express edition" to become a gateway to actual production-capable purchases.

Oracle and IBM need SMB stories to be considered full service. This is something that becomes more critical in a maturing market, such as what some indicators are suggesting we have with database management systems. Both have solid higher-end products. The only alternative to this strategy would seem to be acquiring MySQL. Interestingly, Postgres and Ingres were adopted already, but not to lower-end products, but to the higher-end products of Netezza and Datallegro.

Share:

November 5, 2005

Feeling Hot, Hot, Hot?

According to this recent article in ADTmag.com, citing a survey by Foote Partners, "Data Managers are Hottest of the Hot."

The article says data professionals are in good shape for the next 12 months and only selectively subject to outsourcing or offshoring.

I have subscribed to the Earl Nightingale school of thought, which says you are compensated based on:
1. What you do
2. How well you do it
3. Who you do it for
4. The difficulty of replacing you

According to the article, data management professionals are at least doing something demanding. Now about "how well you do it" and "who you do it for", that is up to you....

Share: