According to the books online http://msdn.microsoft.com/en-us/library/ms177443.aspx
:
"In a clustered index, the leaf nodes contain the data pages of the
underlying table."
Which agrees with your assertion.
From a performance perspective, it DOES work very well. Which is why
I keep hoping for it to show up in PostgreSQL.
On Jul 16, 2009, at 2:21 PM, Scott Carey wrote:
> I could be wrong, but I think MSSQL only keeps the data specified in
> the
> index in the index, and the remaining columns in the data.
> That is, if there is a clustered index on a table on three columns
> out of
> five, those three columns in the index are stored in the index,
> while the
> other two are in a data portion. But it has been several years
> since I
> worked with that DB.
>
> They are certainly storing at least those columns in the index
> itself. And
> that feature does work very well from a performance perspective.
>
> IOT in Oracle is a huge win in some cases, but a bit more clunky for
> others
> than Clustered Indexes in MSSQL. Both are highly useful.
>
> On 7/16/09 10:52 AM, "Justin Pitts" <justinpitts@gmail.com> wrote:
>
>> ISTR that is the approach that MSSQL follows.
>>
>>>
>>> Storing the full tuple in an index and not even having a data only
>>> page
>>> would also be an interesting approach to this (and perhaps simpler
>>> than a
>>> separate index file and data file if trying to keep the data in the
>>> order of
>>> the index).
>>
>>
>