I've "known", if I may abuse the term so, that TSM is capable of using
checksums to inform its incremental decisions for some time.  But it
looks like I've just imagined this; am I correct?

I understand that doing checksum comparisons would be hideously
CPU-expensive at the client side, because you'd basically have to
re-checksum the whole box every incr.  But I've got a client with a
use-case where it would actually be sane to do it, and so I went to
look for the option, and found nada; all the checksum work appears to
be validation of transport and extra paranoia bits for data in a
stgpool.

So was this entirely a hallucination?


- Allen S. Rout