Why Oracle May Really Be Doomed This Time

I don't think Sarah Lacy knows very much about enterprise software sales cycle, or the concerns of enterprise customers.

When you have enterprise customers, their tolerance for downtime is extremely small. When you become big enough, your IT infrastructure becomes a competitive advantage. Maybe more accurately, the C-level executives believe that their IT infrastructure is a competitive advantage, and they will continue to spend money to maintain the status quo. They believe they can't afford to outsource things like CRM, IT Service Management, etc. And enterprise software companies exploit this belief, and extract billions of dollars from them through software upgrades and service contracts.

I worked for enterprise software companies, and I scratched my head wondering why companies kept spending so much money on mediocre software. And it's exactly the old adage from the 80s: "No one gets fired for buying IBM". When you buy Oracle/Siebel, no one will question the status quo. If you throw out Siebel and switch to Salesforce, and Salesforce suffers a 3-day outage much like EC2 did last year, then your head is on the chopping block. That's why CIOs like Oracle.

Now the one thing that Oracle has to really worry about is new companies that never grew up on Oracle, that grow big, like Facebook, Google, etc. The new multi-billion-dollar companies will balk at the idea of spending millions of dollars a year on database licenses, but as long as the enterprise companies are still around, there's more than enough money to sustain Oracle for a long, long time.

A lot of people seem to miss the point about enterprise software and the purveyors of that software; the end goal is not agile or slim (that is incidental) but to enable many, many disparate systems to function at least somewhat coherently. Probably on a level that boggles the minds of most people who have never worked in a large enterprise.

Big businesses end up with lots of disparate technology because they consume a lot of software and it creeps its way into the business by way of developers wanting to play with tools; some C-level moron making 'strategic' decisions; or real demands of the business. But unless you're the technological masters of the universe (like Google) the idea of replacing something wholesale is a recipe for disaster once it's entrenched: it's expensive and what would you replace it with? Who out there can furnish an accounting system that works across 40 different countries that also ties in neatly with a company's logistics or shipping department AND comes with consulting and bespoke development services to handle the 50 billion edge cases that each big business has? This is not a simple problem. This very same thing happens amusingly enough just as frequently in startups and SMEs, but on a different scale ("Oh we couldn't use Mongo, it didn't fit our needs, so we rolled our own -- but then we had to use Frob X and Widget Y and patch it up with Python")

That's why tools like Oracle RDBMS come with such a kitchen sink variety of tools and features.

For instance, I once had to do some reporting and OLAP work across multiple DB systems and stuff the data into an Oracle instance. Unfortunately said systems used a mixture of Oracle, Sybase and MSSQL (oh the wonders of maintaining shrinkwrap products...) so we bought a database link for Sybase and MSSQL so we could query the databases from inside Oracle using two-phase commit. Sure, each adapter cost us $15k but so what? That's pocket change compared to the hackneyed approach of using some external, asynchronous batch job. All this money and patchwork effort in the name of report generation -- but that's all businesses care about.

I feel a lot of the pain he mentions. I remember designing a SOA system many years ago where transaction sharing across web services would save a ton of effort writing the rollback logic that would be needed otherwise. Went to the Oracle web site and saw it supports WS-*, awesome! That must include WS-AtomicTransaction and whatnot. Went to a solution specialist and he said sure you can do transaction sharing! Just export your EJB as a web service and use JTA! Fire up the technology and get things running - not supported at all. Research a bit more - more technical Oracle employees on official forums pointing out it isn't working either. Ugh. They wasted so much of my time by lying and saying they supported something they didn't back then. Definitely not happy with those old crappy systems the article makes fun of every couple paragraphs, nor the whole sales culture where they try to sell me something that doesn't actually fit my needs.

Argh. Pando Daily again.

I don't know how a site can be worse than TechCrunch, but they've managed it.

Oracle is entrenched far too deeply within the inner workings of IT operations of corporations worldwide to ever truly be doomed.