Monday, January 28, 2008

Beyond Performance and Tuning

Now: Performance Management products focus on:
- What query is running ‘bad’ ?
- Do I need to rewrite SQL or add indexes ?
- Is is the response of my entire user community ?
- How many ‘transactions’ can a process in one minute ?

Future: Are the business services that are making my company the most money being completed in a timely fashion ?
- What are my key business services ?
- At month and year end our the business services that make my company the most money getting the resources they need ?
- For this need:
- A tool/product that business people can use to identify their key business processes/transactions
- A product/system that can automatically monitor these key business process and allocate the appropriate middle tier and database resources.

Really all about
Business Relevance not Performance & Tuning

Sunday, January 27, 2008

Beyond Relational


Unlike other industries the IT industry changes at a fast past. The typically life span of enterpirse level technology (consumer technologies change much faster) is 25 years. Take a look at network and Novell; they lasted about 25 years. Take a look at 'green screen' data entry. This last about 25 years until client/server and then web based applications came along. Speaking of client/server and the pace of change. Client/server really only lasted about 10 years before web n-tier based development became the de-facto standard. So what about databases. The relational model has existed for about 15 years (in systems of any significant enterprise caliber). What is going to happen with database storage. Well, like mainframe legacy data stores such as VSAM, Datacom, IDMS, IMS, Adabas etc. they have a very long life span. This is because data is at the heart of the enterprise. You can easily swap out user interface, and in some cases change the network protocol you use. Even though data stores don't change that often and there migration is disruptive to not only your applications, customers and systems but to the entire business, I am seeing a movement to other data storage model more in line with the internet spend of doing business, temporary nature of data and the distributed world to information retrieval and processing. The parameters of this change will involve a data storage model which is:

1. A distributed GRID

2. Data location is transparent to the application

3. Data and code being tightly coopled but dynamically partitioned on the GRID