Mailing-List: contact user-help@uima.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@uima.apache.org
Received-SPF: pass (nike.apache.org: domain of eaepstein@gmail.com designates
 209.85.215.47 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type:content-transfer-encoding;
        b=Dzh/l5udoNHkSu+XQO9yb05urodapG2jITdvu0Lttdl7eh4JykT7weqgqQ/BNT7Hq8
         88rqjVHRgn/c5tVbzTblJ2jBlYf9fR6zPfgCwwZnOJN82pYw0FOgGqM9QW8jYD3dZnql
         mUUUDmAnqD8s6HcknEUbbI0xVjTtkcqyVBhqo=
MIME-Version: 1.0
In-Reply-To: <4C75F813.9010308@orkash.com>
References: <4C6D039D.4070001@orkash.com>
	<AANLkTik0WnG3w-DtBQ_svtAzz551RJm1LdjX+YWcV1_d@mail.gmail.com>
	<4C6E8156.4010600@orkash.com>
	<AANLkTikw2_06zA00z1vgSGnSNzZWupMXP9xTnFD8fDn1@mail.gmail.com>
	<4C75F813.9010308@orkash.com>
Date: Thu, 26 Aug 2010 16:30:23 -0400
Message-ID: <AANLkTikFxrioOKcit9do9eNB2RwAiVO7z_BRYRctj3nR@mail.gmail.com>
Subject: Re: Data read problem in UIMA-AS
From: Eddie Epstein <eaepstein@gmail.com>
To: user@uima.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Lokesh,

Are you planning to run multiple copies of the service? If so, were
you going to distribute the input data ahead of time to the different
machines? If not, the only other reason to use UIMA AS is to optimize
multi-threaded operation on a single machine. Can you say what your
UIMA AS goals are?

Anyway, as you said, data flow can be a performance bottleneck. If it
is, then figure 5 at http://uima.apache.org/doc-uimaas-what.html could
be what you want. If the job is mostly data flow limited, you may want
to look into running the UIMA analytics under Hadoop. There are
references to that on the Apache UIMA web.

Eddie

On Thu, Aug 26, 2010 at 1:13 AM, lokesh chanana
<lokesh.chanana@orkash.com> wrote:
> Hello Eddie,
>
> Once again thanks for replying.
> I tried removing the collection reader.. this thing now works the same wa=
y
> it worked in my CASE 2* .
>
> The main thing I am concern with that in this case I need to send a loads=
 of
> data over the net.. can't there be a way by which I could have rested my
> data on machine hosting the remote service.. and would have called my cli=
ent
> without a collection reader.. or either providing only reference or some
> thing so that my network traffic could be lesser n processing be bit
> faster?? as i would practically never want to download the data first to
> client from my database and then =A0sending it to remote machine..
>
> Secondly, when I define the the collection reader with the remote service
> and it even have data in the local system... what makes the processing go
> extremely slow?? indeed I was expecting it to be more faster... am I wron=
g
> somewhere??
>
> Regards,
> Lokesh