I see. FYI sys(2015) duplication on fast computers are reported (I haven't yet see myself - mine counts as a very slow box these days).
PS: MP boxes might duplicate it as well. However under normal circumstances unlikely.

I wonder how you can have a unique stationID ? Wouldn't it at least require 6 bytes ?
Cetin

>Hi,
>Seek time is not something to worry about, is about the same.
>I choosed the other way because I have three things, the PK, the timestamp of record creation (in sys(2015) and the workstation id, all in one column.
>
>
>>>Hi Cetin
>>>After 100000 inserts
>>>
>>>The index size is 2,994,176 bytes
>>>
>>>After Reindex is 2,161,152 bytes
>>>
>>>I think that is because are random generated.
>>>From what I've read about B-tree indexes, this can be the cause.
>>>
>>>
>>>
>>Hi Dorin,
>>I got a higher value (near 4Mb) but I don't think index size would be my main cause to choose or not to choose it. I wondered seek times. I never tried it myself thinking GUID's unicity is more valuable to me than the nanoseconds gained per seek.
>>Cetin