Re: MSB performance results? #aai #msb #usecaseui #performance

Huabing Zhao <zhaohuabing@...>

Hi Keong, 

Either this Friday or next Monday at 5:00 UTC will work for me.  

Seshu, what time do you prefer if you would like to join the discussion?

On Wed, Apr 24, 2019 at 12:13 PM Keong Lim <Keong.Lim@...> wrote:

Hi Huabing,


Yes, we could have a meeting to discuss it. I would be available on Friday 26th April or Monday 29th April.

Please cc Seshu, as he had some knowledge of the problems in the UUI-to-AAI calls.


According to this

The times for India-China-Australia meeting are limited, but maybe 5.00 UTC is possible?






From: 赵化冰 [mailto:zhaohuabing@...]
Sent: Friday, April 19, 2019 3:52 PM
To: onap-discuss <onap-discuss@...>; Keong Lim <Keong.Lim@...>
Subject: Re: [onap-discuss] MSB performance results? #aai #msb #usecaseui #performance


Hi Keong,


MSB API Gateway's performance testing result is 50,000 QPS on a 8 core machine as long as the back-end service is not the bottleneck.  So it's more likely a restriction from the back-end side.


I suggest we schedule a meeting to discuss this issue and figure out the solution? What do you think?







On Thu, Apr 18, 2019 at 10:02 AM Keong Lim <keong.lim@...> wrote:

Hi all,

I have received reports that the UUI calling to AAI via MSB suffers from some performance problems, apparently stemming from the number of REST API calls required to retrieve data and the fix apparently being to reset the MSB. I know that AAI also has performance problems, but those only result in slowness, not requiring a reset to fix the service.

While I don't have specific figures for the size/number/frequency of the transactions, it seems reasonable that testing the system during development should not be hitting any particular performance limits, e.g. much less than what a production system might experience.

We have suggested some optimisations to consolidate the number of REST calls using, but that is only a temporary reprieve since the amount of data always grows.

So, I would like find out what are the documented performance test results/expectations for MSB? Where are the limitations?

On the wiki, I found:


which only describes some hopeful test methods for Beijing release, but no results shown. On


the suggestion is that S3P performance level 1 has been achieved, but does not link to the evidences page.

Where can I find the evidences pages?


Join to automatically receive all group messages.