Re: [openstack-dev] [Neutron] Data-plane performance testing with Shaker

2015-05-15 Thread Vincent JARDIN

On 15/05/2015 05:09, Kevin Benton wrote:

Even if your bottleneck is the underlay, it still tells you the
performance of the overlay that your tenants will experience, which is
the relevant piece of information for a deployer.


agreed, it makes sense so it create per se a standard definition of 
benchmarking.


Do you have a repo to archive all the reports of benchmarks that have 
been run?


Thank you,

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


[openstack-dev] [Neutron] Data-plane performance testing with Shaker

2015-05-14 Thread Ilya Shakhat
Hi all!

Let me introduce you Shaker - a tool for data-plane performance testing in
OpenStack. The motivation behind it is to have a simple way for measuring
networking bandwidth between instances.

Shaker key features are:

1. *User-defined topology*. The topology is specified as Heat template, so
users may do arbitrary configuration for instances, networks, routers,
floating ips, etc. Instance scheduling is controlled, it is possible to
specify number of instances per compute node and their location.

2. *Simultaneous test execution*. By default Shaker runs tests
synchronously on all deployed instances. It is also possible to increase
the load, thus measuring dependency on number of concurrently working
instances. The feature is useful when one needs to find bottleneck in the
cloud (like usage of non-DVR routers).

3. *Pluggable tools*. Out of the box Shaker supports iperf, netperf and
able to calculate aggregated stats based on their output. Adding a new tool
is easy, in the simplest case it does not even require coding.

4. *Interactive report*. Shaker produces report as single-page HTML
application. The report contains aggregated charts for bandwidth depending
on concurrency, bandwidth per node and precise timeline of traffic on every
node. The report does not have any dependencies and can be shared easily.

If you are interested in knowing more about Shaker welcome to Neutron
Lightning talk presentation by Oleg Bondarev next Wed in Vancouver (
http://sched.co/3BNR).

And certainly welcome to use and contribute!
Code: https://github.com/stackforge/shaker
Docs: http://pyshaker.readthedocs.org/
Launchpad: https://launchpad.net/shaker/
PyPi - https://pypi.python.org/pypi/pyshaker/

Thanks,
Ilya
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [Neutron] Data-plane performance testing with Shaker

2015-05-14 Thread Alan Kavanagh
Hi Ilya

I am interested in this and many thanks for posting this. I have to ask how 
relevant the performance testing is given that Neutron overlays are dependent 
on the underlay? I believe your point 4 below I can see some uses and value 
for, but I am struggling to this been used as a “tool for data-plane 
performance testing” in Neutron networks.

I look forward to the lightning talks.
/Alan

From: Ilya Shakhat [mailto:ishak...@mirantis.com]
Sent: May-14-15 11:30 PM
To: OpenStack Development Mailing List (openstack-dev@lists.openstack.org)
Subject: [openstack-dev] [Neutron] Data-plane performance testing with Shaker

Hi all!
Let me introduce you Shaker - a tool for data-plane performance testing in 
OpenStack. The motivation behind it is to have a simple way for measuring 
networking bandwidth between instances.

Shaker key features are:

1. User-defined topology. The topology is specified as Heat template, so users 
may do arbitrary configuration for instances, networks, routers, floating ips, 
etc. Instance scheduling is controlled, it is possible to specify number of 
instances per compute node and their location.

2. Simultaneous test execution. By default Shaker runs tests synchronously on 
all deployed instances. It is also possible to increase the load, thus 
measuring dependency on number of concurrently working instances. The feature 
is useful when one needs to find bottleneck in the cloud (like usage of non-DVR 
routers).

3. Pluggable tools. Out of the box Shaker supports iperf, netperf and able to 
calculate aggregated stats based on their output. Adding a new tool is easy, in 
the simplest case it does not even require coding.

4. Interactive report. Shaker produces report as single-page HTML application. 
The report contains aggregated charts for bandwidth depending on concurrency, 
bandwidth per node and precise timeline of traffic on every node. The report 
does not have any dependencies and can be shared easily.
If you are interested in knowing more about Shaker welcome to Neutron Lightning 
talk presentation by Oleg Bondarev next Wed in Vancouver (http://sched.co/3BNR).
And certainly welcome to use and contribute!
Code: https://github.com/stackforge/shaker
Docs: http://pyshaker.readthedocs.org/
Launchpad: https://launchpad.net/shaker/
PyPi - https://pypi.python.org/pypi/pyshaker/

Thanks,
Ilya


__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [Neutron] Data-plane performance testing with Shaker

2015-05-14 Thread Kevin Benton
Even if your bottleneck is the underlay, it still tells you the
performance of the overlay that your tenants will experience, which is
the relevant piece of information for a deployer.

On Thu, May 14, 2015 at 3:15 PM, Alan Kavanagh
alan.kavan...@ericsson.com wrote:
 Hi Ilya



 I am interested in this and many thanks for posting this. I have to ask how
 relevant the performance testing is given that Neutron overlays are
 dependent on the underlay? I believe your point 4 below I can see some uses
 and value for, but I am struggling to this been used as a “tool for
 data-plane performance testing” in Neutron networks.



 I look forward to the lightning talks.

 /Alan



 From: Ilya Shakhat [mailto:ishak...@mirantis.com]
 Sent: May-14-15 11:30 PM
 To: OpenStack Development Mailing List (openstack-dev@lists.openstack.org)
 Subject: [openstack-dev] [Neutron] Data-plane performance testing with
 Shaker



 Hi all!

 Let me introduce you Shaker - a tool for data-plane performance testing in
 OpenStack. The motivation behind it is to have a simple way for measuring
 networking bandwidth between instances.

 Shaker key features are:


 1. User-defined topology. The topology is specified as Heat template, so
 users may do arbitrary configuration for instances, networks, routers,
 floating ips, etc. Instance scheduling is controlled, it is possible to
 specify number of instances per compute node and their location.


 2. Simultaneous test execution. By default Shaker runs tests synchronously
 on all deployed instances. It is also possible to increase the load, thus
 measuring dependency on number of concurrently working instances. The
 feature is useful when one needs to find bottleneck in the cloud (like usage
 of non-DVR routers).


 3. Pluggable tools. Out of the box Shaker supports iperf, netperf and able
 to calculate aggregated stats based on their output. Adding a new tool is
 easy, in the simplest case it does not even require coding.


 4. Interactive report. Shaker produces report as single-page HTML
 application. The report contains aggregated charts for bandwidth depending
 on concurrency, bandwidth per node and precise timeline of traffic on every
 node. The report does not have any dependencies and can be shared easily.

 If you are interested in knowing more about Shaker welcome to Neutron
 Lightning talk presentation by Oleg Bondarev next Wed in Vancouver
 (http://sched.co/3BNR).

 And certainly welcome to use and contribute!

 Code: https://github.com/stackforge/shaker

 Docs: http://pyshaker.readthedocs.org/
 Launchpad: https://launchpad.net/shaker/
 PyPi - https://pypi.python.org/pypi/pyshaker/



 Thanks,

 Ilya






 __
 OpenStack Development Mailing List (not for usage questions)
 Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
 http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev




-- 
Kevin Benton

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev