Mailing-List: contact user-help@zookeeper.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@zookeeper.apache.org
Received-SPF: pass (athena.apache.org: domain of baskar.duraikannu@outlook.com
 designates 65.54.190.155 as permitted sender)
Message-ID: <BAY169-W99257095FE66B226623E27EE570@phx.gbl>
Content-Type: multipart/alternative;
	boundary="_924ce822-5614-4665-a501-12d1961f2e8b_"
From: Baskar Duraikannu <baskar.duraikannu@outlook.com>
To: "user@zookeeper.apache.org" <user@zookeeper.apache.org>
Subject: RE: Zookeeper performance
Date: Wed, 31 Jul 2013 18:48:05 -0400
Importance: Normal
In-Reply-To: 
 <CANLc_9KiDsG7HCiua+-tFVOeWzg7GqAPL-i6aaVwgP2mxwZw7w@mail.gmail.com>
References: 
 <BAY403-EAS414CC5EA19C1FA5F7B0CB24EE570@phx.gbl>,<CANLc_9KiDsG7HCiua+-tFVOeWzg7GqAPL-i6aaVwgP2mxwZw7w@mail.gmail.com>
MIME-Version: 1.0

--_924ce822-5614-4665-a501-12d1961f2e8b_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

We are using optimistic concurrency similar to one explained in http://en.w=
ikipedia.org/wiki/Optimistic_concurrency_control
I went through the "version" concept in zookeeper. I do not understand how =
that could be used for "concurrency control". Are you suggesting that all w=
rites go to zookeeper first before making it to the database.=20
We are trying to do it differently... For example=2C let us say we are tryi=
ng to update rowKey 1 of Person table=2C
1. Create a ephemeral node /locks/Person/rowkey12. Zookeeper may create a n=
ode called /locks/Person/rowKey1-0000 (This means that we got lock)3. If we=
 got=2C /locks/Person/rowKey1-0001 =2C we decrement and keep checking child=
 nodes of /locks/Person until  /locks/Person/rowKey1-0001 becomes the first=
 one in the list.=20
Pls help/=20

> From: phunt@apache.org
> Date: Wed=2C 31 Jul 2013 08:47:01 -0700
> Subject: Re: Zookeeper performance
> To: user@zookeeper.apache.org
>=20
> On Wed=2C Jul 31=2C 2013 at 4:16 AM=2C Baskar Duraikannu
> <baskar.duraikannu@outlook.com> wrote:
> >
> > We are looking to use zookeeper for optimistic concurrency. Basically w=
hen the user saves data on a screen=2C we need to lock=2C  read to ensure t=
hat no one else has changed the row while user is editing data=2C persist d=
ata and unlock znode.
> >
>=20
> Why not use the built in optimistic locking that ZK provides? See the
> "version" parameter here:
> http://zookeeper.apache.org/doc/r3.4.5/api/org/apache/zookeeper/ZooKeeper=
.html#getData(java.lang.String=2C
> boolean=2C org.apache.zookeeper.AsyncCallback.DataCallback=2C
> java.lang.Object)
> http://zookeeper.apache.org/doc/r3.4.5/api/org/apache/zookeeper/ZooKeeper=
.html#setData(java.lang.String=2C
> byte[]=2C int=2C org.apache.zookeeper.AsyncCallback.StatCallback=2C
> java.lang.Object)
>=20
> > If the app/thread does not get a lock=2C we may set a watch so that pol=
ling is avoided.
> >
> > Our application is write intensive certain times of the day. We may get=
 about 100k requests per second.  Can zookeeper handle this volume?
>=20
> Unlikely you'll see this level of performance:
> http://zookeeper.apache.org/doc/r3.4.5/zookeeperOver.html#Performance
>=20
> Patrick
 		 	   		  =

--_924ce822-5614-4665-a501-12d1961f2e8b_--