Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (nike.apache.org: domain of saint.ack@gmail.com designates
 74.125.82.51 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:sender:in-reply-to:references:date
         :x-google-sender-auth:message-id:subject:from:to:cc:content-type
         :content-transfer-encoding;
        b=Jt2tPHfAlnZyvE0zQMQ0hT0JF+djvs1T7DFo49GUS9oEtpDqJPkkxrt02Wqozau5Rm
         9B58VMIH4YH2yX7eaGY/WzeSo997jQc9vd2JZsL1ECZACHsysKCMb2baOgAndT6Vvl77
         H+S7f5Pgln6mPz/o7fv5QvjnOBaLoo0u4GW40=
MIME-Version: 1.0
Sender: saint.ack@gmail.com
In-Reply-To: <471587.49260.qm@web130120.mail.mud.yahoo.com>
References: <471587.49260.qm@web130120.mail.mud.yahoo.com>
Date: Thu, 10 Mar 2011 13:47:32 -0800
Message-ID: <AANLkTi=7O8G5L3sEBw+7GozfgP2qC30+YHxumygULeAf@mail.gmail.com>
Subject: Re: A list of HBase backup options
From: Stack <stack@duboce.net>
To: user@hbase.apache.org
Cc: Otis Gospodnetic <otis_gospodnetic@yahoo.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Thu, Mar 10, 2011 at 12:31 PM, Otis Gospodnetic
<otis_gospodnetic@yahoo.com> wrote:
>> Options 1) and 2) will give you a snapshot on a table at a =A0particular
>> instance in time. =A0You'll get the state of the row at the =A0time the
>> MapReduce job crosses that row.
>
> Hm, isn't this contradictory? =A0That is, doesn't "snapshot of a table at=
 a
> particular instance in time" means that I'd get a snapshot of *all* rows =
at a
> single point in time, and not a value of a row when the Export or Copy MR=
 job
> crosses it?
>

Sorry for the sloppy phrasing.

Being a distributed system, getting a consistent view on a table at a
particular moment in time would be a little tough.  The only thing we
guarantee -- currently w/ some caveats, HBASE-2856 -- is a consistent
at the row level only.


> Also, it seems like all options are per-table, right? =A0There is nothing=
 other
> than near real-time full-cluster replication that would back up all table=
s at
> once?

Right.


> This is important when you have multiple tables storing data that depend =
on each
> other. =A0Imagine tables A and B where table B depends on A. =A0If you fi=
rst back up
> A, then by the time I back up B, it may reference some data in A that my =
A's
> backup doesn't contain. =A0If you flip the order and first back up B, the=
n by the
> time I back up A it may contain some extra data that B's backup doesn't r=
efer
> to.
>

Yes.


> Simply put, the backup copies of these 2 tables won't be in sync.
>
> How do people deal with this?
>

If you want them in sync, you are into the world of cross-row,
cross-table transactions.  Replicating the tables will eventually be
consistent with each other (Replication is edit-scoped, not table or
even x-table scoped).

> Would it make sense to document this sort of stuff on
> http://hbase.apache.org/book/book.html ?
>

You mean the list of backup options?  Yes.  And their individual
failings/constraints.

(Otis, in this list you've made other useful 'lists' -- the reporting
one for instance -- that I've put on my 'doc this' list, my list of
things to add into the manual when have a minute).

St.Ack