incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sylvain Lebresne <sylv...@datastax.com>
Subject Re: batch isolation
Date Mon, 04 Jun 2012 08:58:20 GMT
On Sun, Jun 3, 2012 at 6:05 PM, Todd Burruss <bburruss@expedia.com> wrote:
> I just meant there is a "row delete" in the same batch as inserts - all to
> the same column family and key

Then it's the timestamp that will decide what happens. Whatever has a
timestamp lower or equal to the tombstone timestamp will be deleted
(that stands for insert in the batch itself).

--
Sylvain


>
>
> -----Original Message-----
> From: Sylvain Lebresne [sylvain@datastax.com]
> Received: Sunday, 03 Jun 2012, 3:44am
> To: user@cassandra.apache.org [user@cassandra.apache.org]
> Subject: Re: batch isolation
>
> On Sun, Jun 3, 2012 at 2:53 AM, Todd Burruss <bburruss@expedia.com> wrote:>
>> 1 – does this mean that a batch_mutate that first sends a "row delete"
>> mutation on key X, then subsequent insert mutations for key X is isolated?
>
> I'm not sure what you mean by having "a batch_mutate that first sends
> ... then ...", since a batch_mutate is a single API call.
>
>> 2 – does isolation span column families for the same key  within the same
>> batch_mutate?
>
> No, it doesn't span column families (contrarily to atomicity). There
> is more details in
> http://www.datastax.com/dev/blog/row-level-isolation.
>
> --
> Sylvain

Mime
View raw message