Blogger: Lyn Robison
Right now, I have a few things on my radar screen. Those things are concepts, technologies, companies, trailblazing practitioners, and notable failures. The things I am tracking will change over time, but as of right now, here is what I am thinking about -- you might find these helpful in your work.
Concepts
Information Asset Management is a necessary component of competent IT
As a prerequisite for any effort in business intelligence, data center management, compliance audits, electronic discovery, or data security, you must learn what databases you have, what data they contain, and how their data are used. That is called Information Asset Management. Without Information Asset Management, here is what happens:
- Business dissatisfaction with IT, because businesspeople cannot get reliable information to do their jobs.
- Ballooning costs for data storage, backup, server hardware, and data centers.
- Embarrassing, unauthorized disclosures of sensitive information.
- An inability for business decision-makers to properly assess business risk.
- Poor decisions based on inaccurate information.
- The failure of the enterprise.
How important is it for an enterprise to know its data? Just ask some of the financial service firms who suffered because of bad loan data.
I am continually watching tools and vendors that perform database discovery, meta-data discovery, and automated data quality. The right tools for Information Asset Management, along with the right people and the right strategy, are the keys to asset management for enterprise information.
It is better to bridge existing silos than to build more or bigger silos
Silos are ubiquitous in today’s enterprises, and efforts to eliminate them are largely ineffective. Enterprise applications, such as CRM systems, merely replace small silos with big silos, and big silos are not much better than small ones. SOA is an attempt to make silos disappear through process integration, but process integration is not as easy as it looks.
Bridging silos through data integration is an easier path to success. First, you study the important business data in each silo – both the data you have and the data you need – and then identify the instances of the data in each silo. Finally, you reconcile the instances of data between silos. IOW, you figure out, for example, whether “Customer abc123” in one silo is the same as “Customer 47” in another silo. Then you can do cross-silo joins of business information about your customers, which businesspeople will love. This requires nothing but straightforward data management – IT people don’t build more systems, they don’t write more software, they don’t implement more silos, and they don’t chase process integration using SOA.
I am constantly watching for tools and technologies that will bridge existing silos while avoiding the creation of more silos.
Technology
RESTful data services
I am disinclined to bet my career on SOAP and WS-* and process automation/integration using SOA. Instead, I will place my faith in elegant data services that provide data integration and useful information that is both machine- and human-readable across multiple silos.
XML
With XML, you decorate your data with tags that are in essence, declarative code. Once you have done that, you can use hundreds of existing software packages (which your IT department does not have to write) to find, display, edit, save, and process enterprise data. If your data is in XML, you don’t have to write software in order to use it. XML lets you outsource your software and concentrate on your data.
Companies
(this is in no way a blanket endorsement of these companies, rather an acknowledgement that they offer some potentially valuable technology)
Mark Logic
This company has produced the first commercially viable XML database. They are doing for XML databases now what Oracle did for relational databases in the 1980’s.
Masai Technologies
This company has produced tools for automated data visualization, discovery, scanning, analytics, and search. Want to know how many databases you have and what they contain? Tools from Masai Technologies could be very helpful.
IBM
IBM has the most complete vision for holistic data management of any of the major vendors.
Trailblazers
Hewlett-Packard IT
In the past three years, HP’s IT department has consolidated 85 data centers to six; 700 datamarts to fewer than 55; 6,000 applications to about 1,500. Now, 70% of employee time is spent on new development with just 30% on IT support.
Central Intelligence Agency IT
In a presentation at the Enterprise Architecture Conference on October 23, 2007, David Roberts, CIA Chief Technical Officer, explained the CIA’s approach for bridging silos using data integration. It is a brilliant approach that can be implemented gradually. (See the “It is better to bridge existing silos” section above.)
Notable Failures
FBI Virtual Case File (VCF)
The $170 million failure of the VCF should be the poster child for the result you get when you try to replace a whole bunch of little silos with one big silo. (See http://www.usnews.com/usnews/news/articles/050328/28fbi.htm.) The FBI’s failed approach with the VCF is in direct contrast to the CIA’s successful silo-bridging approach.
DBMS-OO-XML Development Stack
Software developers in enterprise IT always use object-oriented programming languages, with a relational database on the backend, and XML for data transfer. The DBMS-OO-XML development stack offers so many choices for data-handling that important pieces of data and their definitions can easily be flung throughout the enterprise in several different locations and formats, or lost altogether. The DBMS-OO-XML development stack is overused and misapplied by most IT development groups.
Reuse
The pinnacle of effectiveness and efficiency that most IT shops try to achieve is reuse – reuse of code through OO programming, reuse of software through components, reuse of application logic through SOA. This is all a big snipe hunt. The real pinnacle of effectiveness and efficiency is not the reuse of software or code, but the retirement of obsolete IT systems and databases. Any IT department that is deliberate about retiring their old systems is way more effective and efficient than those IT departments that are deliberate about reuse. (See the “Hewlett-Packard IT” section above.)
If you would like to suggest other concepts, technologies, companies, trailblazers, and notable failures, I am eager to hear about them.
Comments