[DB-SIG] (slightly off topic) Very Large Database recommendations?

[Sean Davis]

Federico Di Gregorio wrote:
> Il giorno gio, 04/10/2007 alle 11.16 +0200, M.-A. Lemburg ha scritto:
>   
>>> I suppose that python is not really relevant to the question since
>>>       
>> it 
>>     
>>> would be a web database service where the client is python.
>>>
>>> Right now I am interested in getting an idea of how much such a
>>>       
>> system 
>>     
>>> would cost, and where to look for solutions.
>>>
>>> Any ideas where I could look for information relevant to this
>>>       
>> question?
>>
>> The table size doesn't look all that big. Any of Oracle, DB2 or
>> even PostgreSQL should be able to handle it.
>>     
>
> If I remeber correctly I once got an email from a US hospital that was
> using PostgreSQL + psycopg to store medical data and full images into
> the database. They had millions of lines and were about to buy a 6TB SAN
> for the storage. You would be quite safe with 1,000,000 rows and 1TB of
> data.
>   

I'm a postgres user, but I have used Oracle and MySQL as well.  Any of 
these databases on appropriate hardware will deal with VERY large 
datasets (on the order of several TB of data with billions of rows).  As 
the database becomes larger, more hardware becomes necessary (more 
memory, in particular) and it becomes more complicated to manage schema 
to keep the database quick (if that is important), but size is really 
not a limiting factor for most practical purposes.  If you have many 
simultaneous queries, then you will have a slightly different set of 
problems and will need to think about setting up clusters, but only 
careful testing will tell you what is going to be needed in the end. 

Sean