Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1420673 > unrolled thread

[PATCH 0/6] Fast balloon & fast live migration

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/6] Fast balloon & fast live migration Liang Li <liang.z.li@intel.com> - 2016-06-13 12:00 +0200
    RE: [PATCH 0/6] Fast balloon & fast live migration "Li, Liang Z" <liang.z.li@intel.com> - 2016-06-23 10:30 +0200

#1420673 — [PATCH 0/6] Fast balloon & fast live migration

The implementation of the current virtio-balloon is not very
efficient, bellow is test result of time spends on inflating
the balloon to 3GB of a 4GB idle guest:

a. allocating pages (6.5%, 103ms)
b. sending PFNs to host (68.3%, 787ms)
c. address translation (6.1%, 96ms)
d. madvise (19%, 300ms)

It takes about 1577ms for the whole inflating process to complete.
The test shows that the bottle neck is the stage b and stage d.

If using a bitmap to send the page info instead of the PFNs, we
can reduce the overhead in stage b quite a lot. Furthermore, it's
possible to do the address translation and the madvise with a bulk
of pages, instead of the current page per page way, so the overhead
of stage c and stage d can also be reduced a lot.

In addition, we can speed up live migration by skipping process
guest's free pages.

Patch 1 and patch 2 are the kernel side implementation which are
intended to speed up the inflating & deflating process by adding a
new feature to the virtio-balloon device. And now, inflating the
balloon to 3GB of a 4GB idle guest only takes 200ms, it's about 8
times as fast as before.


Patch 3 and patch 4 add the cache drop support, now hypervisor can
request the guest to drop it's cache. It's useful before inflating
the virtio-balloon and before starting live migration.

Patch 5 and patch 6 save guest's free page information into a page
bitmap and send the bitmap to host through balloon's virt queue. 

Liang Li (6):
  virtio-balloon: rework deflate to add page to a list
  virtio-balloon: speed up inflate/deflate process
  mm:split the drop cache operation into a function
  virtio-balloon: add drop cache support
  mm: add the related functions to get free page info
  virtio-balloon: tell host vm's free page info

 drivers/virtio/virtio_balloon.c     | 321 +++++++++++++++++++++++++++++++-----
 fs/drop_caches.c                    |  22 ++-
 include/linux/mm.h                  |   1 +
 include/uapi/linux/virtio_balloon.h |   2 +
 mm/page_alloc.c                     |  40 +++++
 5 files changed, 339 insertions(+), 47 deletions(-)

-- 
1.9.1

[toc] | [next] | [standalone]

#1429570

Any comments?

Liang


> -----Original Message-----
> From: Li, Liang Z
> Sent: Monday, June 13, 2016 5:47 PM
> To: kvm@vger.kernel.org
> Cc: virtio-dev@lists.oasis-open.org; qemu-devel@nongun.org; linux-
> kernel@vger.kernel.org; mst@redhat.com; Li, Liang Z
> Subject: [PATCH 0/6] Fast balloon & fast live migration
> 
> The implementation of the current virtio-balloon is not very efficient, bellow
> is test result of time spends on inflating the balloon to 3GB of a 4GB idle
> guest:
> 
> a. allocating pages (6.5%, 103ms)
> b. sending PFNs to host (68.3%, 787ms)
> c. address translation (6.1%, 96ms)
> d. madvise (19%, 300ms)
> 
> It takes about 1577ms for the whole inflating process to complete.
> The test shows that the bottle neck is the stage b and stage d.
> 
> If using a bitmap to send the page info instead of the PFNs, we can reduce
> the overhead in stage b quite a lot. Furthermore, it's possible to do the
> address translation and the madvise with a bulk of pages, instead of the
> current page per page way, so the overhead of stage c and stage d can also
> be reduced a lot.
> 
> In addition, we can speed up live migration by skipping process guest's free
> pages.
> 
> Patch 1 and patch 2 are the kernel side implementation which are intended
> to speed up the inflating & deflating process by adding a new feature to the
> virtio-balloon device. And now, inflating the balloon to 3GB of a 4GB idle
> guest only takes 200ms, it's about 8 times as fast as before.
> 
> 
> Patch 3 and patch 4 add the cache drop support, now hypervisor can request
> the guest to drop it's cache. It's useful before inflating the virtio-balloon and
> before starting live migration.
> 
> Patch 5 and patch 6 save guest's free page information into a page bitmap
> and send the bitmap to host through balloon's virt queue.
> 
> Liang Li (6):
>   virtio-balloon: rework deflate to add page to a list
>   virtio-balloon: speed up inflate/deflate process
>   mm:split the drop cache operation into a function
>   virtio-balloon: add drop cache support
>   mm: add the related functions to get free page info
>   virtio-balloon: tell host vm's free page info
> 
>  drivers/virtio/virtio_balloon.c     | 321
> +++++++++++++++++++++++++++++++-----
>  fs/drop_caches.c                    |  22 ++-
>  include/linux/mm.h                  |   1 +
>  include/uapi/linux/virtio_balloon.h |   2 +
>  mm/page_alloc.c                     |  40 +++++
>  5 files changed, 339 insertions(+), 47 deletions(-)
> 
> --
> 1.9.1

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web