Cassandra Optimistic Locking

问题:

I have a cassandra table1:

CREATE TABLE Policy.table1 (
name VARCHAR ,
date TIMESTAMP ,
version_num INT,
PRIMARY KEY   (
         name
         )) WITH caching = 'all'
     --   and memtable_flush_period_in_ms = 7200 ;
     ;

I need to implement optimistic locking on tis table. When we read a row from table1 we remember its version_num. And when we want to update this row we compare current version_num value and value that we remembered. Also we need to increment version_num on each update.

Problems:

1. We cannot put version_num into where clause, this will create an error: Bad Request: Non PRIMARY KEY version_num found in where clause:

    update table where name = 'abc' and version = 3
   

2. We cannot set version_num as a part of primary key, because we need to update its value

3. If we index version_num it will not help for update statements, same exception will be thrown
4. The only way I see is to get current version_num value by Java, and if expected and actual version_num values are the same - than execute update. The problem is that in this case we have not atomic operation of checking version_num value and update the row.

Do you see any solution for this problem?

回答1:

The solution was found there: Cassandra 2.0, Lightweight transactions http://www.datastax.com/documentation/cassandra/2.0/cassandra/dml/dml_ltwt_transaction_c.html

In case I execute query:

update table1 set version_num = 5 where name = 'abc'  if version_num = 4;

I will receive a row with [applied] column. This row contains boolean value: true = update was successful, false = in other case.