Return-Path: X-Original-To: archive-asf-public-internal@cust-asf2.ponee.io Delivered-To: archive-asf-public-internal@cust-asf2.ponee.io Received: from cust-asf.ponee.io (cust-asf.ponee.io [163.172.22.183]) by cust-asf2.ponee.io (Postfix) with ESMTP id 57432200D35 for ; Tue, 7 Nov 2017 20:34:58 +0100 (CET) Received: by cust-asf.ponee.io (Postfix) id 543D0160C00; Tue, 7 Nov 2017 19:34:58 +0000 (UTC) Delivered-To: archive-asf-public@cust-asf.ponee.io Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by cust-asf.ponee.io (Postfix) with SMTP id 73200160BEB for ; Tue, 7 Nov 2017 20:34:57 +0100 (CET) Received: (qmail 98160 invoked by uid 500); 7 Nov 2017 19:34:56 -0000 Mailing-List: contact user-help@ignite.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: user@ignite.apache.org Delivered-To: mailing list user@ignite.apache.org Received: (qmail 98151 invoked by uid 99); 7 Nov 2017 19:34:56 -0000 Received: from mail-relay.apache.org (HELO mail-relay.apache.org) (140.211.11.15) by apache.org (qpsmtpd/0.29) with ESMTP; Tue, 07 Nov 2017 19:34:56 +0000 Received: from [192.168.75.66] (c-67-160-238-197.hsd1.ca.comcast.net [67.160.238.197]) by mail-relay.apache.org (ASF Mail Server at mail-relay.apache.org) with ESMTPSA id 307F71A00A5 for ; Tue, 7 Nov 2017 19:34:56 +0000 (UTC) From: Denis Magda Content-Type: multipart/alternative; boundary="Apple-Mail=_5947749E-B4C9-47D7-A070-E3308A2708DC" Mime-Version: 1.0 (Mac OS X Mail 10.3 \(3273\)) Subject: Re: Data region's data page size Date: Tue, 7 Nov 2017 11:34:55 -0800 References: <9b45cf4d-f1be-16b9-3aa4-02dc39efbfe1@gmail.com> To: user@ignite.apache.org In-Reply-To: Message-Id: X-Mailer: Apple Mail (2.3273) archived-at: Tue, 07 Nov 2017 19:34:58 -0000 --Apple-Mail=_5947749E-B4C9-47D7-A070-E3308A2708DC Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 Yes, you should if average object size exceeds 4KB because otherwise = Ignite will spend more time spreading and maintaining a single object = across several pages.=20 =E2=80=94 Denis > On Nov 7, 2017, at 11:14 AM, Andrey Kornev = wrote: >=20 > Thanks Ivan! A follow-up question, if you don't mind.=20 >=20 > Should I consider changing the default 4K page size to 8K (or possibly = to 16K) if the average size of my cache objects is, say, 12K, with max = size up to 100K? >=20 > Thanks > Andrey >=20 > From: Ivan Rakov > Sent: Tuesday, November 7, 2017 8:22 AM > To: user@ignite.apache.org > Subject: Re: Data region's data page size > =20 > Hello Andrey, >=20 > Short answer: just keep default 4k page size, it will work fine unless = you have very specific hardware. >=20 > Too small page size causes write amplification on hardware level: = https://www.cactus-tech.com/resources/blog/details/write-amplification-as-= seen-in-solid-state-drives = > On the other hand, too big page size causes write amplification on = Ignite level. Whole 16k page will be persisted if you have changed only = few bytes. >=20 > You may also want to check our durable memory tuning guide for more = details:https://apacheignite.readme.io/docs/durable-memory-tuning = > Best Regards, > Ivan Rakov >=20 > On 07.11.2017 1:28, Andrey Kornev wrote: >> Hello, >>=20 >> What are the considerations for choosing the best data page size? It = looks like there are only four possible values (1K, 4K, 8K and 16K) to = choose from, and I don't immediately see why one would want to use = anything other than 16K? >>=20 >> Is it because the value lookup within a page is linear and therefore, = I want to keep the number of values that fit on a single page = sufficiently low? >>=20 >> Please advise! >>=20 >> Thanks >> Andrey --Apple-Mail=_5947749E-B4C9-47D7-A070-E3308A2708DC Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 Yes, you should if average object size exceeds 4KB because = otherwise Ignite will spend more time spreading and maintaining a single = object across several pages. 

=E2=80=94
Denis

On Nov 7, 2017, at 11:14 AM, = Andrey Kornev <andrewkornev@hotmail.com> wrote:

Thanks Ivan! A follow-up question, if you don't = mind. 

Should I consider changing the default 4K page size to 8K (or = possibly to 16K) if the average size of my cache objects = is, say, 12K, with max size up to 100K?

Thanks
Andrey

From: Ivan Rakov <ivan.glukos@gmail.com>
Sent: Tuesday, November 7, 2017 = 8:22 AM
To: user@ignite.apache.org
Subject: Re: Data region's data page = size
 
Hello Andrey,

Short answer: just keep default 4k page size, it will work = fine unless you have very specific hardware.

Too small page size causes write amplification on hardware = level: https://www.cactus-tech.com/resources/= blog/details/write-amplification-as-seen-in-solid-state-drives
On the = other hand, too big page size causes write amplification on Ignite = level. Whole 16k page will be persisted if you have changed only few = bytes.

You may also want to check our durable memory tuning = guide for more details:https:/= /apacheignite.readme.io/docs/durable-memory-tuning
Best Regards,
Ivan Rakov

On = 07.11.2017 1:28, Andrey Kornev wrote:
Hello,

What are the = considerations for choosing the best data page size? It looks like = there are only four possible values (1K, 4K, 8K and 16K) to choose = from, and I don't immediately see why one would want to use = anything other than 16K?

Is it because the value = lookup within a page is linear and therefore, I want to keep = the number of values that fit on a single page sufficiently = low?

Please advise!

Thanks
Andrey

= --Apple-Mail=_5947749E-B4C9-47D7-A070-E3308A2708DC--