My bad, I meant exactly that, there could be performance issues if you have 
multiple nodes hitting the same row since that row work items are serialised.

Dean

From: aaron morton <aa...@thelastpickle.com<mailto:aa...@thelastpickle.com>>
Reply-To: "user@cassandra.apache.org<mailto:user@cassandra.apache.org>" 
<user@cassandra.apache.org<mailto:user@cassandra.apache.org>>
Date: Thursday, January 3, 2013 12:28 PM
To: "user@cassandra.apache.org<mailto:user@cassandra.apache.org>" 
<user@cassandra.apache.org<mailto:user@cassandra.apache.org>>
Subject: Re: Multi threads updating single row

Multiple nodes could be a problem
Not sure what you mean here Dean.

There are no issues with multiple clients, from multiple threads, processes or 
nodes inserting / updating to the same row.

There are potential performance issues though due to row level isolation used 
in the write path. The first (internal) writing thread "wins" when updating a 
row, so any concurrent (internal) write threads updating the same row must 
(again internally) restart their processing.

You will probably only see issues when you have 10's of clients continually 
updating the same row. And then the size of  issue depends somewhat on the size 
of the inserts you are running. i.e. a smaller insert size means less re-work.

There are also potential performance issues if you have secondary indexes on 
the row.

Hope that helps.

-----------------
Aaron Morton
Freelance Cassandra Developer
New Zealand

@aaronmorton
http://www.thelastpickle.com

On 4/01/2013, at 2:32 AM, "Hiller, Dean" 
<dean.hil...@nrel.gov<mailto:dean.hil...@nrel.gov>> wrote:

Multiple nodes could be a problem but multiple threads is probably just fine.  
If you have two threads write to the same column, the last one wins though so I 
hope your timestamps are unique even across threads so you don't lose data ;).

Dean

From: Jay Svc 
<jaytechg...@gmail.com<mailto:jaytechg...@gmail.com><mailto:jaytechg...@gmail.com>>
Reply-To: 
"user@cassandra.apache.org<mailto:user@cassandra.apache.org><mailto:user@cassandra.apache.org>"
 
<user@cassandra.apache.org<mailto:user@cassandra.apache.org><mailto:user@cassandra.apache.org>>
Date: Wednesday, January 2, 2013 4:55 PM
To: 
"user@cassandra.apache.org<mailto:user@cassandra.apache.org><mailto:user@cassandra.apache.org>"
 
<user@cassandra.apache.org<mailto:user@cassandra.apache.org><mailto:user@cassandra.apache.org>>
Subject: Multi threads updating single row

Happy New Year Everyone..!

In the situation, there are multiple client thread updating or adding new 
columns in a same row. These new columns is a time series data.

My Question is - Would multiple threads able to add columns to the same row? Do 
you see any performance issues since there is a comparator and has to sort 
columns as they get added. Do you see any loss of data or conflict due to 
mulitple thread updating data? I am expecting high volumn of data coming on 
each one of those threads.

Thanks in advance.

Jay


Reply via email to