I am working on a python 3.6 based script to monitor dns resolution time, using dns python. I need to understand what is the best approach to get the resolution time. I am currently using the following:
Method:1
import dns.resolver
import time
dns_start = time.perf_counter()
answers = dns.resolver.query(domain, qtype)
dns_end = time.perf_counter()
print("DNS record is: " ,answers.rrset)
print('DNS time =' ,(dns_end - dns_start)* 1000,"ms")
Method 2
import dns.resolver
import time
answers = dns.resolver.query(domain, qtype)
print("DNS record is: " ,answers.rrset)
print('DNS time =' , answers.response.time * 1000,"ms")
Thanks
"It depends" as the famous saying says.
You need to define what is "best" for you in "best approach". The two methods do things correctly, just different things.
First one will count both network time, remote server processing, and then local generation and parsing of DNS messages, including time spent by Python in the dnspython library.
The second method takes only into account the network and remote processing time: if you search for response_time in https://github.com/rthalley/dnspython/blob/master/dns/query.py you will see it is computed as a difference between the time just before sending the message (hence after it is been build, so the generation time will not be included) and just after it has been received (hence before it is parsed, so parsing time will not be included).
Second method can basically test the remote server performances, irrespective of your local program (including Python itself and the dnspython library) performances (but the network time needed between you and remote server will be counted in).
First method shows the perceived time, as it includes all time needed in the code to do something, that is build the DNS packet and parse the result, actions without which the calling application could do nothing with the DNS content.