Reply to: Db table with most records you have worked with

[PAH]

Sounds like a lot of people have very inefficient data storage methods.

There's only so many unique letters and numbers.

[ThomasSoerensen]

My planB db reached 190gb at the last customer analysis - and that is without the transaction log - had to delete a lot of other data on the server to make room for it.

[NotAllThere]

I once worked with a VB developer, interfacing his application to SAP. He boasted that his database was approaching 1MB in size. At some time I casually pointed out that the GL table alone was 31GB.

[Lockhouse]

One of my systems has a database with c1.5 billion records, all small records though. Looking at it is on my list but as it's working well at the moment it's not a priority.

[doodab]

Not to mention more profitable for you, even if you worked as the janitor.

A US President visited NASA one day, saw a janitor working feverishly at sweeping the floor, asked him
what he was doing and received the reply, "Because I'm working to put a man on the moon"

So to extend sas' example you would be working to put pointless bulltulip on the internet. Doesn't have quite the same ring to it.

[AtW]

Not to mention more profitable for you, even if you worked as the janitor.

A US President visited NASA one day, saw a janitor working feverishly at sweeping the floor, asked him
what he was doing and received the reply, "Because I'm working to put a man on the moon"

[sasguru]

I wish I worked at Twitter, life would have been so much easier.

HTH

Not to mention more profitable for you, even if you worked as the janitor.

[AtW]

Updating database of 66,555,636,804 rows every 8 hours.

The largest I've worked with is 3,529,437,390,545 rows.

I wish I worked at Twitter, life would have been so much easier.

HTH

[ChrisPackit]

What is your record?

My record is 42 rows.

Hope this helps.

[NotAllThere]

There's a load into the Data Warehouse of a few hundred million records every year - fortunately only a handful of fields and it is being summarised. The source, however, never gets trimmed down and must hold several billion records of ~50 fields.

I've heard the health insurance companies here are already running databases in the Petabyte range.

[minestrone]

Oracle used something similar for early versions of oracle workflow. Oh how we laughed when the people replacing our "pilot" system decided to model 4000 attributes directly (including such niceties as site_1_address, site_2_address, site_3_address, site_4_address) rather than using the order number as a route into a proper data model. 100,000 orders, 400 million rows (this was on fairly low end sun hardware, our system ran happily on a sun ultra something with 8 disks IIRC) and slightly less than stellar interactive performance. What's that Bob, you need to go back to the drawing board? Do you want to borrow my crayons?

There was a great drive in the industry against hard coding anything for a few years to the point where objects and tables were completely abstracted away from being what they represent. Stick users, accounts, addresses, transaction into one table of type attribute data columns.

Nobody had a clue what the software does on these systems, you cannot look at the code and work out what is going to happen because it is all run time based.

Sticking a break point on setName was never an option when a name attribute got fecked up, you have to stick a break point on setAttribute and drift through countless cycles of the method.

[PAH]

The table holding the complete known list of prime numbers must be getting pretty large.

At least it only requires one column.

Largest known prime number - Wikipedia, the free encyclopedia

The record passed one million digits in 1999, earning a $50,000 prize.[4] In 2008 the record passed ten million digits, earning a $100,000 prize.[5] Additional prizes are being offered for the first prime number found with at least one hundred million digits and the first with at least one billion digits.

I spot a plan B. Anyone done a p2p prime number generator app yet?

[doodab]

I worked on a schema which for a major retail bank which consisted of a 3 table model of type, attribute & data. Oracle had to come in when performance somehow unsurprisingly struggled and claimed it was the biggest table size they had ever seen.

Oracle used something similar for early versions of oracle workflow. Oh how we laughed when the people replacing our "pilot" system decided to model 4000 attributes directly (including such niceties as site_1_address, site_2_address, site_3_address, site_4_address) rather than using the order number as a route into a proper data model. 100,000 orders, 400 million rows (this was on fairly low end sun hardware, our system ran happily on a sun ultra something with 8 disks IIRC) and slightly less than stellar interactive performance. What's that Bob, you need to go back to the drawing board? Do you want to borrow my crayons?

[minestrone]

I worked on a schema for a major retail bank which consisted of a 3 table model of type, attribute & data. Oracle had to come in when performance somehow unsurprisingly struggled and claimed it was the biggest table size they had ever seen.

[doodab]

About 4 billion, although it was a materialised view consisting of a 4 way cartesian self join and a couple of other little tables. I did try and warn them it wasn't a good idea.