Return-Path: X-Original-To: apmail-hadoop-hdfs-user-archive@minotaur.apache.org Delivered-To: apmail-hadoop-hdfs-user-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 8A4ABDA69 for ; Mon, 13 Aug 2012 15:54:28 +0000 (UTC) Received: (qmail 64414 invoked by uid 500); 13 Aug 2012 15:54:23 -0000 Delivered-To: apmail-hadoop-hdfs-user-archive@hadoop.apache.org Received: (qmail 64317 invoked by uid 500); 13 Aug 2012 15:54:23 -0000 Mailing-List: contact user-help@hadoop.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: user@hadoop.apache.org Delivered-To: mailing list user@hadoop.apache.org Received: (qmail 64310 invoked by uid 99); 13 Aug 2012 15:54:23 -0000 Received: from athena.apache.org (HELO athena.apache.org) (140.211.11.136) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 13 Aug 2012 15:54:23 +0000 X-ASF-Spam-Status: No, hits=2.5 required=5.0 tests=FREEMAIL_REPLY,FSL_RCVD_USER,HTML_MESSAGE,RCVD_IN_DNSWL_LOW,SPF_PASS X-Spam-Check-By: apache.org Received-SPF: pass (athena.apache.org: domain of matthias.zengler@gmail.com designates 209.85.212.48 as permitted sender) Received: from [209.85.212.48] (HELO mail-vb0-f48.google.com) (209.85.212.48) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 13 Aug 2012 15:54:16 +0000 Received: by vbme21 with SMTP id e21so4439255vbm.35 for ; Mon, 13 Aug 2012 08:53:56 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:content-type; bh=HlBRPqsfrOA0MoRt1JLESvHoeU8e9XJaFbr6ThxHwiI=; b=LXdbfC83MEkNUxa/4ZXTKEt6bAzuXVm2iCB5w06DSOoxNgMywKTv82050FarVfTXDl 1PGN3my6WvbazQO3OXVY+7m4ruUV6KVwyv+VDGjDBUYWjvexQwVxiZnNfgAUgq7C6+yz TJQxNF/FH7Plm/JFSykMHc256liGiDFaFmSvlpuEQZ50TBfmb38V3ESJ9g7L4sN29LNK 3n2ZhpxzIaO4NbgNs5WbP1sqlHQb8tAJqdqCxEaeeimZLKBS9MNOjvrfBKNSeW7s1ux+ YMeKwps8B1w/GWA9LCC+jNbfL0BdVCf1+hTqakXblxC35ZcjVux/iiVEuKBGOxGpZ2ZE +P9w== MIME-Version: 1.0 Received: by 10.58.31.228 with SMTP id d4mr9666429vei.40.1344873235869; Mon, 13 Aug 2012 08:53:55 -0700 (PDT) Sender: matthias.zengler@gmail.com Received: by 10.58.226.196 with HTTP; Mon, 13 Aug 2012 08:53:55 -0700 (PDT) In-Reply-To: References: <1430515874-1344869393-cardhu_decombobulator_blackberry.rim.net-742405440-@b2.c15.bise7.blackberry> Date: Mon, 13 Aug 2012 17:53:55 +0200 X-Google-Sender-Auth: elTLshkLaAsbOKJbq1m4PlqH8DI Message-ID: Subject: Re: how to enhance job start up speed? From: Matthias Kricke To: user@hadoop.apache.org Content-Type: multipart/alternative; boundary=047d7b41c3ea7defb604c727b39f X-Virus-Checked: Checked by ClamAV on apache.org --047d7b41c3ea7defb604c727b39f Content-Type: text/plain; charset=ISO-8859-1 @Bejoy KS: Thanks for your advice. @Bertrand: It is parallelisable, this is just a test case. In later cases there will be a lot of big files which should be processed completly each in one map step. We want to minimize the overhead of network traffic. The idea is to execute some worker (could be different stuff, e.g. wordcount, linecount, translation etc) at the node where the file is situated. If I get it right so far, we need to do several things... first chunk size should be as big as the file. Then the file is on a single node of the hadoop cluster, am I right? And set the file to non splitable. Did you have some more advice? Anyway thanks so far! Greetings, MK 2012/8/13 Bertrand Dechoux > It was almost what I was getting at but I was not sure about your problem. > Basically, Hadoop is only adding overhead due to the way your job is > constructed. > Now the question is : why do you need a single mapper? Is your need truly > not 'parallelisable'? > > Bertrand > > > On Mon, Aug 13, 2012 at 4:49 PM, Bejoy KS wrote: > >> ** >> Hi Matthais >> >> When an mapreduce program is being used there are some extra steps like >> checking for input and output dir, calclulating input splits, JT assigning >> TT for executing the task etc. >> >> If your file is non splittable , then one map task per file will be >> generated irrespective of the number of hdfs blocks. Now some blocks will >> be in a different node than the node where map task is executed so time >> will be spend here on the network transfer. >> >> In your case MR would be a overhead as your file is non splittable hence >> no parallelism and also there is an overhead of copying blocks to the map >> task node. >> Regards >> Bejoy KS >> >> Sent from handheld, please excuse typos. >> ------------------------------ >> *From: * Matthias Kricke >> *Sender: * matthias.zengler@gmail.com >> *Date: *Mon, 13 Aug 2012 16:33:06 +0200 >> *To: * >> *ReplyTo: * user@hadoop.apache.org >> *Subject: *Re: how to enhance job start up speed? >> >> Ok, I try to clarify: >> >> 1) The worker is the logic inside my mapper and the same for both cases. >> 2) I have two cases. In the first one I use hadoop to execute my worker >> and in a second one, I execute my worker without hadoop (simple read of the >> file). >> Now I measured, for both cases, the time the worker and >> the surroundings need (so i have two values for each case). The worker took >> the same time in both cases for the same input (this is expected). But the >> surroundings took 17% more time when using hadoop. >> 3) ~ 3GB. >> >> I want to know how to reduce this difference and where they come from. >> I hope that helped? If not, feel free to ask again :) >> >> Greetings, >> MK >> >> P.S. just for your information, I did the same test with hypertable as >> well. >> I got: >> * worker without anything: 15% overhead >> * worker with hadoop: 32% overhead >> * worker with hypertable: 53% overhead >> Remark: overhead was measured in comparison to the worker. e.g. >> hypertable uses 53% of the whole process time, while worker uses 47%. >> >> 2012/8/13 Bertrand Dechoux >> >>> I am not sure to understand and I guess I am not the only one. >>> >>> 1) What's a worker in your context? Only the logic inside your Mapper or >>> something else? >>> 2) You should clarify your cases. You seem to have two cases but both >>> are in overhead so I am assuming there is a baseline? Hadoop vs sequential, >>> so sequential is not Hadoop? >>> 3) What are the size of the file? >>> >>> Bertrand >>> >>> >>> On Mon, Aug 13, 2012 at 1:51 PM, Matthias Kricke < >>> matthias.mk.kricke@gmail.com> wrote: >>> >>>> Hello all, >>>> >>>> I'm using CDH3u3. >>>> If I want to process one File, set to non splitable hadoop starts one >>>> Mapper and no Reducer (thats ok for this test scenario). The Mapper >>>> goes through a configuration step where some variables for the worker >>>> inside the mapper are initialized. >>>> Now the Mapper gives me K,V-pairs, which are lines of an input file. I >>>> process the V with the worker. >>>> >>>> When I compare the run time of hadoop to the run time of the same >>>> process in sequentiell manner, I get: >>>> >>>> worker time --> same in both cases >>>> >>>> case: mapper --> overhead of ~32% to the worker process (same for >>>> bigger chunk size) >>>> case: sequentiell --> overhead of ~15% to the worker process >>>> >>>> It shouldn't be that much slower, because of non splitable, the mapper >>>> will be executed where the data is saved by HDFS, won't it? >>>> Where did those 17% go? How to reduce this? Did hadoop needs the whole >>>> time for reading or streaming the data out of HDFS? >>>> >>>> I would appreciate your help, >>>> >>>> Greetings >>>> mk >>>> >>>> >>> >>> >>> -- >>> Bertrand Dechoux >>> >> >> > > > -- > Bertrand Dechoux > --047d7b41c3ea7defb604c727b39f Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable
@Bejoy KS: Thanks for your advice.

@Bertrand: It i= s parallelisable, this is just a test case. In later cases there will be a = lot of big files which should be processed completly each in one map step. = We want to minimize the overhead of network traffic. The idea is to execute= some worker (could be different stuff, e.g. wordcount, linecount, translat= ion etc) at the node where the file is situated.=A0

If I get it right so far, we need to do several things... fi= rst chunk size should be as big as the file. Then the file is on a single n= ode of the hadoop cluster, am I right? And=A0
set the file to non= splitable.

Did you have some more advice? Anyway thanks so far!

Greetings,
MK

2012/8/13 Bertrand Dechoux <dechouxb@gmail.com>
It was almost what I was getting at but I wa= s not sure about your problem.
Basically, Hadoop is only adding overhea= d due to the way your job is constructed.
Now the question is : why do you need a single mapper? Is your need truly n= ot 'parallelisable'?

Bertrand


On Mon, Aug 13, 2012 at 4:49 PM, Bejoy KS <= ;bejoy.hadoop@g= mail.com> wrote:
Hi Matthais

When an mapreduce program is being used ther= e are some extra steps like checking for input and output dir, calclulating= input splits, JT assigning TT for executing the task etc.

If your f= ile is non splittable , then one map task per file will be generated irresp= ective of the number of hdfs blocks. Now some blocks will be in a different= node than the node where map task is executed so time will be spend here o= n the network transfer.

In your case MR would be a overhead as your file is non splittable henc= e no parallelism and also there is an overhead of copying blocks to the map= task node.
Regards
Bejoy KS

Sent from handheld, please = excuse typos.
From: Matthias Kricke <matthias.mk.kricke@gmail.com>
Date: Mon, 13 Aug 2012 16:33:06 +0200
Subject: Re: how to enhance job start up speed?

Ok, I try to clarify:

1) The worker is the logi= c inside my mapper and the same for both cases.
2) I have two cases. In= the first one I use hadoop to execute my worker and in a second one, I exe= cute my worker without hadoop (simple read of the file).
=A0 =A0Now I measured, for both cases, the time the worker and the=A0s= urroundings=A0need (so i have two values for each case). The worker took th= e same time in both cases for the same input (this is expected). But the su= rroundings took 17% =A0more time when using hadoop.
3) ~ =A03GB.

I want to know how to reduce thi= s difference and where they come from.=A0
I hope that helped? If = not, feel free to ask again :)

Greetings,
MK

P.S. just for your information, I did the same = test with hypertable as well.=A0
I got:=A0
=A0* worker = without anything: 15% overhead
=A0* worker with hadoop: 32% overh= ead
=A0* worker with hypertable: 53% overhead
Remark: overhead w= as measured in comparison to the worker. e.g. hypertable uses 53% of the wh= ole process time, while worker uses 47%.

2012/8/13 Bertrand Dechoux <dechouxb@gmail.com>
I am not sure to understand and I guess I am not the only one.

1) Wh= at's a worker in your context? Only the logic inside your Mapper or som= ething else?
2) You should clarify your cases. You seem to have two case= s but both are in overhead so I am assuming there is a baseline? Hadoop vs = sequential, so sequential is not Hadoop?
3) What are the size of the file?

Bertrand


On Mon, Aug 13, 2012 at 1:51 PM, Matthias Kricke <matthias.mk.kricke@gmail.com> wrote:
Hello all,

I'm using CDH3u3.
If I want to process on= e File, set to non splitable hadoop starts one Mapper and no Reducer (thats= ok for this test scenario). The Mapper
goes through a configurat= ion step where some variables for the worker inside the mapper are initiali= zed.
Now the Mapper gives me K,V-pairs, which are lines of an input file. I= process the V with the worker.

When I compare the= run time of hadoop to the run time of the same process in sequentiell mann= er, I get:

worker time --> same in both cases

case: mapper --> overhead of ~32% to the worker process (same fo= r bigger chunk size)
case: sequentiell --> overhead of ~15% to= the worker process

It shouldn't be that much slower, because of non sp= litable, the mapper will be executed where the data is saved by HDFS, won&#= 39;t it?
Where did those 17% go? How to reduce this? Did hadoop n= eeds the whole time for reading or streaming the data out of HDFS?

I would appreciate your help,

= Greetings
mk




--
Bertrand Dechoux




= --
Bertrand Dechoux

--047d7b41c3ea7defb604c727b39f--