cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <>
Subject Re: handling deletes
Date Mon, 30 Mar 2009 23:53:36 GMT
On Mon, Mar 30, 2009 at 5:37 PM, Jeremy Dunck <> wrote:
> It also assumes that the number of un-GC'able SSTables is small (as
> compared to free disk space), right?

right, but that's not that big of an assumption, since the tombstone
is going to be smaller than whatever value was deleted.  Best to just
document that the operations insert,remove is going to temporarily not
free up all the space as not doing the insert at all.  (this is
actually already the case if you include the commit log.)

> If nodes A and B kept a (key,
> timestamp, tombstone) triple outside of the SSTable

that's a really huge if, since (a) you'd add a whole new layer of
complexity, and (b) it's not going to magically take any less space
outside than inside.


View raw message