Tuesday, November 21, 2006

The Russian Drinking Contest

It's a story I've forced Tuomas Pystenen (Oakie of Deepbase, Finland) to tell again and again.

So his Finnish collegues on a project many moons ago were invited to a party with their Russian project counter-parts. Tuomas was the only one not too much drunk when the Finns arrived in Russia by train, so he was picked by their Russian hosts for the drinking contest.

The idea was brilliant: A Finnish guy and a Russian guy having to each empty six glasses of either water or vodka in front of an audience. If the audience saw any flicker or other nervousness in their faces they had lost. One of the glasses would be vodka, the five others water.

So Tuomas emptied the first glass with 100 grams of liquid in it. It was vodka, but his face didn't move, of course. He's from Finland. So he was happy: The vodka glass had been identified - the rest would be easy.

Time for the second 100 grams glass - it was warm to the touch so Tuomas was certain it was water. It wasn't. It was vodka, and warm vodka goes very fast to the brain. Just so you know.

In fact there was vodka in all the glasses. It was rather hard according to Tuomas and the others who participated. I seem to remember that Tuomas won the prize. Highly deserved.

Monday, November 20, 2006

How CIA stole Oracle from the Russians

Dear readers,

At the UKOUG conference in Birmingham last week, the OakTable got a new member - Kurt van Meerbeeck from Belgium - the author of DUDE, Pepi and Pretoria, and a guy with a slight hint of humour.

His presentation had people laughing rather hard, and so I've chosen to publish his theory on Oracle's origin - in other words, how CIA stole the source code from the Russians, who had developed it originally!

Here we go, in Kurt's own words:

I believe, to truely understand X$ tables you need to have some basic knowledge of an eastern european language ... just to even pronounce them. Now - Oracle used to be named after a CIA project right ?

And Oracle hasexisted ... for how long now... 31 years ?

I think the Americans stole the source code from the Russians during the cold war - and asked Larry and his team to read in the tape (they were developing tape driver software weren't they).

Probably they were expecting some list of agents and I can imaging the CIA was quite dissapointed when they saw this source code full of X$ garble.

I guess they told Larry he could keep the tapes and on he went revealing this relational database power... of course they couldn't use the X$ stuff so they wrote V$ views on top of them ... so people could understand...

It's not a coincidence that after the fall of the Berlin wall a *lot* of features were added - Oracle 6 was almost a complete rewrite !!!

So to end this - and you are probably wondering about this ... how come Jonathan Lewis is such a wiz in X$ tables ???

There's only one explanation - MI6 ! His codename was probably OO-600 ... with a license to kill your instance ;-) (math-teacher ... yeah - right !)

Perhaps it should be pronounced 006-hundred?

Tuesday, November 07, 2006

Funny, very internal, remarks from the OakTable

I know, I know. The internal mailing list we use in the OakTable Network is for exactly that - internal communications.

But now and then it's hard not to let others know what goes on there. This is such a case. I will not name names, and I will not disclose the real reason for these brilliant submissions - but I hope you'll enjoy it.

Here's the first entry:

While I understand your point here, I have to say that using high performance technology to increase the performance of a system where the architecture is broken is a suboptimal approach.

When teaching design classes (design? does database design training still exist, but I digress) I used an example of going shopping to show the benefits of using the array interface:

If I'm going to the store for milk, butter, bread and cheese I can either drive to the store, buy milk, drive home, put the milk in the refrigerator, drive back to the
store, buy butter, drive home, put the butter in the refrigerator etc, or can drive to the store, pick up a basket, put the milk, butter, bread, cheese and beer in the basket, buy everything in the basket, drive home, make a cheese sandwich and drink a beer while congratulating myself for having optimized the shopping task.

Here's the second entry (in response, of course):

DB corruption:

I drive to the store for milk, butter and cheese. I pick up the milk at the back of the store and see that there is a special on beer. Put milk, butter and cheese on the floor of aisle 7, and load basket with beer. On making way to checkout counter, decide that basket is too heavy, so stop in aisle 3 to have a couple of the beers. Find some crisps in aisle 3 - grab some of them and munch them down. Now thirsty again, down 4 more beers from basket in aisle 3. Remember something about needing to come to the shop to get some milk and cheese. Head down to the milk section - have a couple of beers on the way. Oh look, som dufus has left some milk, cheese and butter on the floor in aisle 7. Celebrate with another brewsky. Start singing my favourite football song, washed down with another ale. Vaguely recall something about chilk and meese. Ah what the hell - where's the beer aisle.


That's all, folks. I hope you enjoyed it as much as I did :-).


Saturday, November 04, 2006

M2M and Oracle Legacy

Money-to-Miracle (M2M) is my tongue-in-cheek way of saying that you might need assistance from consultants, thereby spending money, thereby giving my fantastic company money.

So I predicted in the original YPDNR paper that you would need (expensive and specialised) consultants if you were going to use RAC.

Was that true? And if yes, is it still true?

RAC was introduced in Oracle9i, making it RAC version 1.0. Conventional wisdom has it that you should attempt not to run any software in version 1.0.

Fortunately, RAC was based solidly on OPS, which was partly put into version 5 (five!) of Oracle and officially introduced in version 6.0.35 a.k.a. version 6.2, and so it worked pretty much as planned from the beginning. But you certainly needed special skills setting it up and, to a lesser extent, running it. Those skills were hard to come by back then. So yes, M2M.

With the RAC implementations of Oracle10g (especially 10gR2) many things have become much easier, including installation and administration. Also, due to the perhaps 1500 RAC installations worldwide , there are now more RAC-trained people than ever before.

However, something happened: The Oracle DBA world has become a legacy world. It’s the same (older and older) people meeting each other at the conferences and writing to each other in magazines and on mailing lists.

No new, fresh faces show up. Our children want to write game software or design mobile phones instead of becoming Mainframe tape operators or Oracle DBA’s.

Also, the old DBA’s are beginning to either leave the field because they see a brighter future somewhere else, or because they retire or are being diluted, working with other databases, managing application servers and handling storage issues as well as being an Oracle DBA.

Combine that with the surge in IT-spending we’re experiencing at the time of writing (2006), and you can perhaps see the irony in companies hiring super experts from Miracle for ordinary DBA tasks. They cannot find the qualified people anymore.

There’s money in Legacy for years to come, make no mistake about that, and we’re talking a lot of money and we’re talking a lot of years. Just stay on top, stay focused, and don’t get scared about all those other people around you learning all sorts of new programming languages and databases.

Take pride in the fact that Oracle is still by far the best database on the market. Just as the IBM Mainframe or the AS400 are still by far the most stable IT platforms on the market, or the fact that VMS is a fantastic operating system.

So perhaps Oracle is the mainframe of databases, and perhaps PL/SQL is the Cobol of databases – but you won’t find a more stable, scalable, tuneable and widely used platform for database-centric solutions.

According to my Second Law, Oracle will, therefore, be replaced with something less stable and performing. That happens to all technologies that reach a state of perfection.

So you probably don’t need the same amount of specialised consulting skills for setting up and running RAC. But the ordinary skills just got harder to find. Which is why I quote the famous Vera Lynn song to new RAC customers:

We'll meet again,
Don't know where,
Don't know when,
But I know
We'll meet again
Some sunny day.

Wednesday, November 01, 2006

A Brief Intro to Marketing (or: You Need RAC!)

Three and a half years ago I wrote the paper You Probably Don't Need RAC ("YPDNR I"), and that got a few people out of their comfortable chairs.

Most of the official responses from Oracle to 'YPDNR I' have been highly emotional (i.e. negative) despite the fact that I ended the paper with the words “Now prove me wrong”, and offered to correct any mistakes in my paper immediately.

I think there are two reasons for the negative reactions despite lack of any technical arguments against my paper so far.

1. RAC is something Oracle is (justly) technically proud of, and since it’s a unique technical feature it is also a unique selling feature.

2. I used the word NOT in the title.

In marketing terms, using a negative word when talking about your own product is just not done. You always put a positive spin on it:

* Forty-two Reasons for Choosing RAC
* Using RAC in Critical Environments for Better High Availability and Lower Costs

If I had used one of these (or similar) titles, yet had kept the original content, I don’t think anybody would have reacted.

In fact, I’m certain nobody had objected if I had not used NOT in the title, because I’ve seen several colleagues do exactly that: Put up a positive-sounding title of a presentation or paper, and then proceed to document that in fact most people don’t need RAC.

Why would you deliberately choose a positive-sounding title for a presentation about why people probably don’t need a certain thing? I can think of two reasons:

1. So The Big Vendor doesn’t get mad at you.
2. The paper will get more easily accepted by the paper-selection committees of the various user groups (see my previous post about prostitution).

The first reason is widespread: People are afraid of “Making The Big Vendor mad”. Well, companies don’t get mad. People inside companies get mad (or at least react as if they do).

They get mad instead of getting factual because of lack of, well, facts, but that just makes them madder. Feeling powerless is not good.

The second reason (getting papers selected) is a real problem, and certainly not just in the Oracle world. Many user groups these days suck up to The Big Vendor, imposing a ‘self-restraint’ or, to put it in plain English, self-censorship that would make Yahoo and Google (and their Chinese friends) proud.

People keep telling me it is simply because of the money involved. Perhaps. I doubt it, but let's consider it.

If you make fun of Carlsberg’s 'Probably the Best Beer in The World' campaign, you effectively question a multi-billion Kroner marketing investment. If you question RAC, you potentially hurt the Oracle sales force’s forecast for up-selling RAC and other Enterprise Edition (extra-cost) options.

On the other hand, very very few people are actually driven by money.

I say that as someone who studied Economics for a few years, is a liberal at heart, and all that.

All serious research I've seen in the last ten years or more point to the fact that people are driven by other things than financial rewards. Which is just part of the reason why micro-economic theory (the theories, often with very heavy math involved, about actions and interactions of the individual agents in the economy) can't explain anything after all these years, except when they partner up with sociologists and other socialists :-) .

Most of our community is not driven by money. A few individuals are, but you know they are, and it's just an illness they were born with or forced to adopt because of some slip in their childhood.

We shouldn't generalise based on a few exceptions.

But we all want acknowledgement and friends - to beloong to a group, etc. The threat from someone (be that a marketing guy from The Big Vendor) of losing friends or connections or membership of a group can make a lot of people blink.

Just remember: As soon as you blink in situations like that, there's no going back. You have moved a step or two down the ladder. The only thing that might save you is the fact that The Big Vendor will do his annual re-org no matter what and so perhaps move the Pressure Guy somewhere else and replace him with someone who doesn't know you blinked :-) .

I knew these ritualised, annual re-orgs were good for something!