Arno Brinkman wrote:

>
>> ...64bit transaction ids. The 32bit id is good for 136
>>transactions per second for a year, but ... [we are]
>>beginning to approach the point where that is a plausible load.
>
>
> Are you considering this for V3 or V3+ ?
> If we're going to change the ODS structure for the transaction-id we should review it for all stored
> numbers IMO. This can go on with the "new" record-structure as Jim proposed some time ago.

Another problem is page numbers...

Probably 3+, though we should put in code sooner to detect that the
TID is about to rollover and keep it from trashing the database.


Regards,


Ann