Mailing-List: contact user-help@zookeeper.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@zookeeper.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
Message-ID: <28027.78123.qm@web65515.mail.ac4.yahoo.com>
Date: Sun, 30 Jan 2011 16:56:05 -0800 (PST)
From: Andrew Purtell <apurtell@apache.org>
Reply-To: apurtell@apache.org
Subject: Re: Extracting Zab from Zookeeper
To: dev@zookeeper.apache.org
Cc: user@zookeeper.apache.org
In-Reply-To: <4D430A95.1080000@yahoo-inc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

FWIW, we have this crazy idea of using something like an extracted ZAB for =
maintaining quorum-cliques in a modified HBase cluster: https://issues.apac=
he.org/jira/browse/HBASE-2357=0A=0A   - Andy=0A=0A> From: Benjamin Reed <br=
eed@yahoo-inc.com>=0A> Subject: Re: Extracting Zab from Zookeeper=0A> To: "=
dev@zookeeper.apache.org" <dev@zookeeper.apache.org>=0A> Cc: "Ted Dunning" =
<ted.dunning@gmail.com>, "user@zookeeper.apache.org" <user@zookeeper.apache=
.org>=0A> Date: Friday, January 28, 2011, 10:27 AM=0A> flavio and ted are c=
orrect. zab and=0A> zookeeper are tightly coupled, and =0A> as flavio point=
s out, zab provides a bit more than just=0A> atomic broadcast.=0A> =0A> i t=
hink it would be a good thing to do for three reasons:=0A> =0A> 1) it would=
 make testing, benchmarking, and fixing that layer =0A> (especially for ZOO=
KEEPER-22) much easier.=0A> 2) we could try different backends easier. (i'd=
 love to try=0A> using bookkeeper for example.)=0A> 3) although the zab int=
erface must be richer than normal=0A> atomic broadcast (access to past tran=
sactions, for example), i=0A> think in practice applications that do primar=
y/backup can benefit=0A> from this richer interface.=0A> =0A> ben=0A> =0A> =
On 01/28/2011 07:03 AM, Ted Dunning wrote:=0A> > I think that what Flavio w=
as saying is that it is like=0A> > pulling a string on a sweater.=A0 Almost=
 any application that=0A> > wants ZAB is probably going to need the broadca=
st and they the=0A> > broadcast, they will want to have logging.=A0 And tra=
nsactions.=0A> >=A0 And so on.=0A> >=0A> > Moreover, some of these features=
 are not necessarily=0A> > encoded in ZK in a way that you can point to.=A0=
 Instead they are=0A> > enabled by the way that several functional units ar=
e glued=0A> > together.=0A=0A=0A=0A