[Pdns-users] PowerDNS Backend MySQL QPS

[Kenneth Marshall]

On Tue, Jul 19, 2016 at 03:41:37AM +0700, Genzo Rey wrote:
> Dear All,
> 
> I'm looking for new Nameserver solutions for my company. I build 1 VPS
> PowerDNS 512 MB RAM and 1 VPS MySQL 512MB RAM to test performance (qps).
> This is my results:
> 
> [Status] Sending queries (to 192.168.10.143)
> [Status] Started at: Tue Jul 19 02:58:04 2016
> [Status] Stopping after 1 run through file
> [Status] Testing complete (end of file)
> 
> Statistics:
> 
>   Queries sent:         100000
>   Queries completed:    100000 (100.00%)
>   Queries lost:         0 (0.00%)
> 
>   Response codes:       NOERROR 57019 (57.02%), NXDOMAIN 42981 (42.98%)
>   Average packet size:  request 35, response 66
>   Run time (s):         46.909183
>   Queries per second:   2131.778761
> 
>   Average Latency (s):  0.046852 (min 0.002626, max 0.079728)
>  Latency StdDev (s):   0.002579
> 
>  
> 
> I thinks this qps is too low. Can you tell me how to boost performance of my
> Nameserver (pdns.conf, increase RAM - CPU, .). Can you show me what is most
> important in PowerDNS (CPU, RAM, HDD or SSD, .)
> 
> I'm just a newbie and I need more help to build our Nameserver. Thanks very
> much for any help and sorry so much because my poor English.
> 
>  
> 
> Thanks & Best Regards,
> 

Hi,

Your test of performance is not a realistic evaluation of PDS's ability
to serve DNS zones compared to Bind. What you have actually done is to
benchmark the query performance of the backend data stores. Try running
an appropriate test using real zone data and an actual query load. I think
you will find that performance will be impacted greatly by the fact that
most authoritative DNS servers have cached frequently accessed data and
that the performance of the backend data store is not usually an issue.
You might also want to use the Bind backend in PDNS to get a more apples-
to-apples comparison.

Regards,
Ken