Hello,

It seems the performance issue with the switft ceilometer middleware is not new, it was raised before:
https://bugs.launchpad.net/ceilometer/+bug/1337761
Which was abandoned with the Reason: The swift middleware is the site of much need for change, and this change in particular is pretty minor compared to those other needs.

And discussions regarding the fate of the swift middleware. It is quite long:
http://lists.openstack.org/pipermail/openstack-dev/2014-July/041522.html

It seems that the swift middleware will have to change quite a bit in the future to address the performance issue, and it could take a while and won't make it in Icehouse 2014.1.3.

I think Eoghan Glynn should be consulted on this matter as well, since he is a ceilometer core and he participated in the discussions and should be able to tell if this commit should be accepted as a FFE.

Still, IMO, a 10-20% performance drop is still better than not working.

I say it's worthy.

+1

From: Adam Gandelman [adamg@ubuntu.com]
Sent: Monday, September 29, 2014 11:39 PM
Cc: openstack-stable-maint@lists.openstack.org
Subject: Re: [Openstack-stable-maint] Icehouse freeze exception request

Any other stable-maint folks have have opinions on this FFE?

On Sun, Sep 28, 2014 at 9:18 PM, Osanai, Hisashi <osanai.hisashi@jp.fujitsu.com> wrote:

Hi Adam,

Thank you for your response.

On Saturday, September 27, 2014 2:50 AM, Adam Gandelman wrote:
> but would like to know if the performance issues you observed are also present in Juno?

I don't have juno env to measure the performance now but I think it should be observed in
juno from the code point of view.

> If use of the swift middleware is completely unusable in current state, the performance
> hit seems justified.

I think so too.

Thanks again!
Hisashi Osanai