Systems manager Sean Lincolne doesn't live by the motto: "If it ain't broke, don't fix it." When his IT team at...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
the Australian Fisheries Management Authority (AFMA) needed to upgrade its database management system (DBMS), they decided to break away from Microsoft Windows NT servers because NT had broken the AFMA database again and again. Lincolne's team builds and supports heterogeneous IT systems that enable AFMA to manage fishery resources within the 200-nautical mile Australian fishing zone (AFZ) and -- in some cases -- beyond.
In part one of this two-part interview, Lincolne describes why NT was the wrong choice for AFMA's DBMS, details future plans for Linux migrations and explains why AFMA isn't an all-Linux shop. In part two, he offers a step-by-step, in-depth account of his Linux/DBMS migration and its return on investment.
Could you set the scene, outlining the situation that led to the data center makeover?
Sean Lincolne: Back in 1999/2000, AFMA was preparing to upgrade its Ingres [DBMS] from release 6.4 to OpenIngres 1.2. Computer Associates [was withdrawing support] for Ingres release 6.4. Also, we projected increasing demands for performance due to increasing use by end users.
At this stage, our existing HP D320 multi-processor servers [running Microsoft Windows NT and Ingres] were beginning to show their age, and a series of costly hardware upgrades were on the horizon. Ongoing maintenance costs were also a concern.
What problems were you having, and what was running on NT in situations where problems occurred?
Lincolne: The problems we were experiencing were the classic 'blue screen' that plagued NT in the early days. [We had] 'blue screen of death,' multiple patch applications, DLL version conflicts, the usual things that plague NT administrators.
Typically, we would find the machine dead in the morning, and have to reset it and then perform a database recovery.
Unfortunately, this system was also running a vendor-supplied application on top of an Ingres database. Since then, we changed to an alternative vendor-supplied turnkey system and, with additional patching and tweaking, things are much better.
What options did you evaluate, and why did you choose the path you took?
Lincolne: There had been several favorable reviews of vendors' experiences porting their products to Linux. Informix was one that piqued my interest. Linux was gaining credibility as a business platform (rather than a hobbyist toy) and name companies were testing the market with demonstration and beta products.
We became aware of Computer Associates' beta program for its OpenIngres (DBMS) 1.2 release. The fact that such a large player was entering the market, and the stability and performance of the beta, meant that this was an opportunity to exploit.
Furthermore, industry support for Linux and Unix meant that there was a supply of competent people, products and documentation. Being located in Canberra [Australia's capital] meant that there were sufficiently mature and cost-effective support providers available to assist us should the need arise.
Was this AFMA's first look at Linux?
Lincolne: AFMA had been using Linux for other applications since 1994. While these were back-end systems with minimal exposure to end users -- remote access servers, network management, time server, Web proxy, etc. -- I found that their reliability was excellent. One server continually operated for more than two years with no downtime before we had to shut it down to move [to new] buildings.
What other technology factors influenced your decision?
Lincolne: We were also looking at the use of dynamic content on the Internet for both public and private systems. The best direction at the time -- and one that still proves to be a safe choice -- was Apache and PHP. While both Apache and PHP are available on many platforms, including Win32, Linux was well supported by both providers.
So, what did your short list of decision-related pluses and minuses look like?
Lincolne: The main drivers for moving the DBMS technology to Linux were:
-- Bad experiences with Windows NT (patching, server reboots, database corruption, etc.).
-- Good experiences with Linux (stable, high availability, rich selection of technology, availability of external support).
For more information:
Read this SearchEnterpriseLinux.com exclusive: A Linux enterprise with a view: On Red Hat, Novell and more