From: "Michael R. Hines" <mrhines@linux.vnet.ibm.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: aliguori@us.ibm.com, "Michael S. Tsirkin" <mst@redhat.com>,
qemu-devel@nongnu.org, owasserm@redhat.com, abali@us.ibm.com,
mrhines@us.ibm.com, gokul@us.ibm.com
Subject: Re: [Qemu-devel] [RFC PATCH RDMA support v1: 10/13] introduce new command migrate_check_for_zero
Date: Thu, 11 Apr 2013 09:24:57 -0400 [thread overview]
Message-ID: <5166B9A9.9070904@linux.vnet.ibm.com> (raw)
In-Reply-To: <51667FEE.903@redhat.com>
That's very accurate. Zero page scanning *after* the bulk phase
is not very helpful in general.
Are we proposing to skip is_dup_page() after the bulk phase
has finished?
The testcase I'm using is a "worst-case" stress memory hog command
(apt-get install stress) - but against this does not affect anything until
we assume the bulk phase has already completed.
On 04/11/2013 05:18 AM, Paolo Bonzini wrote:
> Il 11/04/2013 09:38, Michael S. Tsirkin ha scritto:
>> On Wed, Apr 10, 2013 at 06:28:18PM -0400, mrhines@linux.vnet.ibm.com wrote:
>>> From: "Michael R. Hines" <mrhines@us.ibm.com>
>>>
>>> This allows the user to disable zero page checking during migration
>>>
>>> Signed-off-by: Michael R. Hines <mrhines@us.ibm.com>
>> IMO this knob is too low level to expose to management.
>> Why not disable this automatically when migrating with rdma?
> Thinking more about it, I'm not sure why it is important to disable it.
>
> As observed earlier:
>
> 1) non-zero pages typically have a non-zero word in the first 32 bytes,
> as measured by Peter Lieven, so the cost of is_dup_page can be ignored
> for non-zero pages.
>
> 2) all-zero pages typically change little, so they are rare after the
> bulk phase where all memory is sent once to the destination.
>
> Hence, the cost of is_dup_page can be ignored after the bulk phase. In
> the bulk phase, checking for zero pages it may be expensive and lower
> throughput, sure, but what matters for convergence is throughput and
> latency _after_ the bulk phase.
>
> At least this is the theory. mrhines, what testcase were you using? If
> it is an idle guest, it is not a realistic one and the decreased
> latency/throughput does not really matter.
>
> Paolo
next prev parent reply other threads:[~2013-04-11 13:25 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-04-10 22:28 [Qemu-devel] [RFC PATCH RDMA support v7: 00/13] rdma cleanup and reordering mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 01/13] introduce qemu_ram_foreach_block() mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 02/13] Core RMDA logic mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 03/13] RDMA is enabled by default per the usual ./configure testing mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 04/13] update QEMUFileOps with new hooks mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 05/13] accessor function prototypes for new QEMUFileOps hooks mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 06/13] implementation of " mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 07/13] introduce capability for dynamic chunk registration mrhines
2013-04-11 2:24 ` Eric Blake
2013-04-11 2:39 ` Michael R. Hines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 08/13] default chunk registration to true mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 09/13] parse QMP string for new 'rdma' protocol mrhines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 10/13] introduce new command migrate_check_for_zero mrhines
2013-04-11 2:26 ` Eric Blake
2013-04-11 2:39 ` Michael R. Hines
2013-04-11 7:52 ` Orit Wasserman
2013-04-11 12:30 ` Eric Blake
2013-04-11 12:36 ` Orit Wasserman
2013-04-11 17:53 ` Michael R. Hines
2013-04-11 3:11 ` Michael R. Hines
2013-04-11 7:38 ` Michael S. Tsirkin
2013-04-11 9:18 ` Paolo Bonzini
2013-04-11 11:13 ` Michael S. Tsirkin
2013-04-11 13:19 ` Michael R. Hines
2013-04-11 13:51 ` Michael S. Tsirkin
2013-04-11 14:06 ` Michael R. Hines
2013-04-11 14:17 ` Paolo Bonzini
2013-04-11 14:35 ` Michael R. Hines
2013-04-11 14:45 ` Paolo Bonzini
2013-04-11 15:37 ` Michael R. Hines
2013-04-11 13:24 ` Michael R. Hines [this message]
2013-04-11 14:15 ` Paolo Bonzini
2013-04-11 14:45 ` Michael S. Tsirkin
2013-04-11 14:57 ` Michael R. Hines
2013-04-11 15:01 ` Michael S. Tsirkin
2013-04-11 15:08 ` Paolo Bonzini
2013-04-11 15:35 ` Michael R. Hines
2013-04-11 15:45 ` Paolo Bonzini
2013-04-11 16:02 ` Michael R. Hines
2013-04-11 16:12 ` Paolo Bonzini
2013-04-11 16:07 ` Eric Blake
2013-04-11 16:29 ` Michael R. Hines
2013-04-11 16:36 ` Eric Blake
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 11/13] send pc.ram over RDMA mrhines
2013-04-11 6:26 ` Paolo Bonzini
2013-04-11 12:41 ` Michael R. Hines
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 12/13] updated protocol documentation mrhines
2013-04-11 2:43 ` Eric Blake
2013-04-11 2:47 ` Michael R. Hines
2013-04-11 6:29 ` Paolo Bonzini
2013-04-10 22:28 ` [Qemu-devel] [RFC PATCH RDMA support v1: 13/13] print out migration throughput while debugging mrhines
2013-04-10 22:32 ` [Qemu-devel] [RFC PATCH RDMA support v7: 00/13] rdma cleanup and reordering Michael R. Hines
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5166B9A9.9070904@linux.vnet.ibm.com \
--to=mrhines@linux.vnet.ibm.com \
--cc=abali@us.ibm.com \
--cc=aliguori@us.ibm.com \
--cc=gokul@us.ibm.com \
--cc=mrhines@us.ibm.com \
--cc=mst@redhat.com \
--cc=owasserm@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).