Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Message-ID: <4BF0E9F3.6020106@apache.org>
Date: Mon, 17 May 2010 00:02:11 -0700
From: Patrick Hunt <phunt@apache.org>
User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US;
 rv:1.9.1.9) Gecko/20100317 Thunderbird/3.0.4
MIME-Version: 1.0
To: user@cassandra.apache.org
CC: Chris Goffinet <cg@chrisgoffinet.com>
Subject: Re: zookeeper, how do you feed the pets?
References: <AANLkTik2zo9yW666KKKU-NhZJZaOyCmHgFHktrlejGQz@mail.gmail.com>
 <2CA40A44-6F19-472D-B335-203298F9F190@chrisgoffinet.com>
In-Reply-To: <2CA40A44-6F19-472D-B335-203298F9F190@chrisgoffinet.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi, ZK uses a quorum protocol (similar but not the same as paxos) for 
writes, as a result it's sensitive to inter-server latency. (however 
reads are always local and therefore not effected) Running a cluster 
fully w/in a colo you can achieve 15k writes/second, with a cluster 
distributed across a wan the write performance will be lower. You can 
see some detail on in-colo performance here http://bit.ly/4B8gC7 and 
here http://bit.ly/4ekN8G

Regards,

Patrick

On 05/16/2010 07:10 PM, Chris Goffinet wrote:
> If you are running multiple datacenters, intend to have a lot of writes for counters, I highly advise against it. We got rid of ZK because of that.
>
> -Chris
>
> On May 16, 2010, at 7:04 PM, S Ahmed wrote:
>
>> Can someone quickly go over how you go about using zookeeper if you want to store counts and have those counts be accurate?
>>
>> e.g. in digg's case I believe, they are using zookeeper so they can keep track of digg's for a particular digg story.
>>
>> Is it a backend change only and then storing API calls are uneffected?  is it a config issue ?
>>
>> What are the ramifications of using this addon, are writes slower because you have to wait for the write to propogate to all the servers?
>