Mailing-List: contact user-help@ignite.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@ignite.apache.org
MIME-Version: 1.0
In-Reply-To: <1453901675908-2745.post@n6.nabble.com>
References: <1453753163474-2706.post@n6.nabble.com>
 <1453757025088-2708.post@n6.nabble.com>
 <1453792131288-2715.post@n6.nabble.com>
 <1453792759683-2716.post@n6.nabble.com>
 <1453796396982-2718.post@n6.nabble.com>
 <1453835664319-2727.post@n6.nabble.com>
 <1453878938319-2734.post@n6.nabble.com> <56A8B138.7090708@gridgain.com>
 <1453898306808-2742.post@n6.nabble.com> <56A8C18B.5090003@gridgain.com>
 <1453901675908-2745.post@n6.nabble.com>
From: Dmitriy Setrakyan <dsetrakyan@apache.org>
Date: Wed, 27 Jan 2016 15:24:26 -0800
Message-ID: 
 <CA+0=VoXOJRQb=O2E0jJ7w6D23kePARxkWGTHHyFOBvf_mXRO9w@mail.gmail.com>
Subject: Re: Sql query performance with partitioned caches
To: user <user@ignite.apache.org>
Content-Type: multipart/alternative; boundary=001a11c33afcb40be8052a591c7b

--001a11c33afcb40be8052a591c7b
Content-Type: text/plain; charset=UTF-8

On Wed, Jan 27, 2016 at 5:34 AM, Andrey Nestrogaev <
a.nestrogaev@flexsoft.com> wrote:

> Denis,
>
> I don't agree that extra newtork roundtreeps "costs almost nothing" for
> heavy OLTP load :)
>

I disagree, because many large companies, including some very well known
banks, use Ignite exactly for OLTP workloads. If you look at the data grid
market, then you will quickly realize that no other data grid vendor or
project comes even close to the SQL capabilities provided by Ignite.

Again, Ignite already supports non-collocated SQL (will be released in
version 1.6 as part of IGNITE-1232
<https://issues.apache.org/jira/browse/IGNITE-1232>), but the performance
of the non-collocated joins may be worse than for the collocated joins
depending on the result sets.

In your design, you need to aim to achieve about +50% collocation for your
queries. Generally, the more collocated queries you have, the better the
performance will be. However, it is OK to have a smaller number of queries,
or the queries that are not executed often, as non-collocated queries.

D.

--001a11c33afcb40be8052a591c7b
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><div class=3D"gmail_quote">=
On Wed, Jan 27, 2016 at 5:34 AM, Andrey Nestrogaev <span dir=3D"ltr">&lt;<a=
 href=3D"mailto:a.nestrogaev@flexsoft.com" target=3D"_blank">a.nestrogaev@f=
lexsoft.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" styl=
e=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Denis,<=
br>
<br>
I don&#39;t agree that extra newtork roundtreeps &quot;costs almost nothing=
&quot; for<br>
heavy OLTP load :)<br></blockquote><div><br></div><div>I disagree, because =
many large companies, including some very well known banks, use Ignite exac=
tly for OLTP workloads. If you look at the data grid market, then you will =
quickly realize that no other data grid vendor or project comes even close =
to the SQL capabilities provided by Ignite.</div><div><br></div><div>Again,=
 Ignite already supports non-collocated SQL (will be released in version 1.=
6 as part of <a href=3D"https://issues.apache.org/jira/browse/IGNITE-1232">=
IGNITE-1232</a>), but the performance of the non-collocated joins may be wo=
rse than for the collocated joins depending on the result sets.</div><div><=
br></div><div>In your design, you need to aim to achieve about +50% colloca=
tion for your queries. Generally, the more collocated queries you have, the=
 better the performance will be. However, it is OK to have a smaller number=
 of queries, or the queries that are not executed often, as non-collocated =
queries.</div><div><br></div><div>D.</div></div></div></div>

--001a11c33afcb40be8052a591c7b--