Its hard not to agree with Dr. DBA when hes so fired up with Oracle Database 10g, telling people the specific time to upgrade to the new database version is, quite simply, “now.”
At the opening keynote at the International Oracle Users Group Live 2004 conference up in Toronto on Monday, Vice President of Product Strategy Ken Jacobs—for the Oracle-unwashed, thats Dr. DBAs alter ego—promised that major enterprise applications such as those of PeopleSoft Inc., SAP AG and Oracle are on the brink of being certified with 10g.
To have such major enterprise applications ready to go in such a short time—Oracle Database 10g was released only in January—is testimony to the level of commitment to 10g that exists on the part of software vendors. It also flattens a large stumbling block to mass adoption. After all, no business can upgrade if its key enterprise applications wont run on the database.
Jacobs and other voices on the “upgrade soon” side of the argument gave me some compelling reasons to upgrade, such as pointing out that support for 8i is ending at years end. Why go to 9i, they said, when you can skip directly to 10g and avoid the cost of two migrations? Also, the first patch set is due out in June. Waiting for minor bugs to be ironed out in major new releases is yet another cause of adoption lag. With the bugs out of the way, plus major applications (almost) ready to go, why wait?
So, I sipped the 10g Kool-Aid, and I admit, it was pretty sweet. After all, we want to believe that 10gs promises are imminent. Shared resources, automatic load balancing, lower total cost of ownership, harnessing of underutilized computing resources, automatic management: Whats not to lust after?
Caution Rules IOUG Live
But the arguments on the other side, which call for the standard, cautious, wait-and-see, stay-away-from-dot-zero-releases attitude, are more compelling. They won the day, Im told, on the IOUG Live panel—run by Ian Abramson, director of education for IOUG—that closed out the conference. The panel was titled, “10g: Upgrade Sooner or Later … That Is the Question!”
I got two of the “later” panel members on the phone. Earlier in the week, they came up with a good acronym to describe their stance. Theyre calling themselves TCM: The Cautious Majority. Here are their very compelling arguments for going slowly:
Skepticism on the promised apps delivery dates.
Oracle is saying that Oracle applications are targeted to be certified on 10g by June. PeopleSoft applications are targeted for summer, and SAP is looking at Q1 2005.
Look at the fine print, though: Those applications are “targeted” for those release dates, not guaranteed. Maybe were a cynical bunch, but delivery-date slippage is too common not to anticipate here.
The patch sets arent worth much until the stuffs proved in production.
There just arent enough sites running this software in production to viably, thoroughly test it. As Carl Olofson, research director at International Data Corp., said in a recent TechTarget newsletter article (sorry, theres no link to give you, but if youre curious to see the article, e-mail me and Ill forward the newsletter), 10g probably wont show up in sales “for at least six months,” since they like to wait for a maintenance agreement to be in place.
As one of the pro-go-slow panel members pointed out to me, its just not worth the risk of migrating if theres a potential for stability issues. “We completely agree that this version of the database may have been the most tested version,” said Gaja Krishna Vaidyanatha, an independent consultant in San Francisco.
“But we heard the same for Version 8 and Version 9. It may be true theyre increasing testing, but why would you subject your enterprise to upgrading to a new release if theres no driving business factor?”
The homogeneity problem.
I talked to Mike Herald, a consultant with Pronto North America, in Eden Prairie, Minn., and another TCM panel member. As he pointed out, there are no big shops that dont have a mix of this and that, whether youre talking databases, operating systems or hardware. Everybodys running heterogeneous shops.
But with Oracle, if you start out with one flavor, youre stuck. The migration path isnt there. If you start out on Unix, youve got to stay on Unix. Same thing with Linux. Plus, its all got to be done with Oracle databases. And just forget trying to run an NT operating system. After all, as Herald put it and as we all know, Larry wants Bill to go away. End of story.
Whats the compelling business value?
This is the main crux of the TCM platforms argument. As Vaidyanatha puts it, if youre an Amazon.com, youve likely got an aggressive upgrade timeline. One concrete example of a compelling business reason to upgrade is exemplified by Heralds wife, who recently deployed 10g in order to remain consistent with Oracle Developments path for Java application deployment, which would enable effective integration with the Oracle customer relationship management (CRM) applications at a later date.
She was working on an application for gathering and configuring end-customer order requirements. Each order for a specific end customer was highly customized, making order information-gathering in the current manual system a very complex and error-prone process. The project, which rang in for about $1 million, deployed using Oracle 10g and following Oracles CRM standards for Java application deployment.
10g has been very successful, Herald reports, in its addressing of the issues at hand, whereas 9i wouldnt allow the proper use of Enterprise JavaBeans (EJB) without implementing a multitude of bug fixes.
No compelling reason to use EJB? No compelling need to rush to an untried platform? Dont drink the Kool-Aid. Join us, The Cautious Majority, and take the tried-and-true, wait-and-see stance toward 10g.
Write to Lisa Vaas at firstname.lastname@example.org.
eWEEK.com Database Center Editor Lisa Vaas has written about enterprise applications since 1997.