Information Session

1
Information Session DNS Service level
recommendations and experiences
2

• What is a DNS Service Level (or SLA)?
• SLAs are a measure of
• The ability to resolve names in the name space
• The ability to obtain authoritative answers for
  domain delegations
• The time taken for changes in registry data to be
  reflected in the name servers

3
How to measure DNS Service Levels? These service
levels are measured in a number of ways. Each has
its own metric and recommended level and each
deals with vastly different areas that effect DNS
availability.
4

• Service Level Measures
• Overall Availability
• Per Name Server Availability
• Round Trip Time - RTT
• Packet Loss
• Average Resolution Time
• Scheduled Down Time
• DNS updates

5
Service Level Measures - Overall
Availability This refers to the ability of an
authoritative answer for DNS requests. Recommende
d SLA 100
6
Service Level Measures Per Name Server
Availability This refers to the availability of
an authoritative answer for a DNS request from
each name server individually. Recommended SLA
99.99
7
Service Level Measures Round Trip Time Packet
Loss This refers to the ping time and number
of packets that go missing in the network that
the name server is connected to. Recommended
SLA 300ms/10
8
Service Level Measures Average Resolution
Times This refers to the average response time
for DNS queries from the time the query is
received to the time the response is
sent. Recommended SLA 250ms
9
Service Level Measures Scheduled downtime This
refers to the amount of scheduled controlled
total DNS downtime allowed. Recommended SLA
No downtime
10
Service Level Measures DNS updates This refers
to the addition, removal and modification of
records in the name space, 5-10min updates are
not uncommon, but now days they are average, with
providers able to offer instant DNS updates, and
propagation using IXFR. Our Recommended SLA
20mins Max
11

• Recommended Service Levels Summary
• 100 Overall DNS Availability
• 99.99 Per Name Server
• 300ms RTT (round trip time)
• 10 Packet Loss maximum
• 250ms average resolution time
• No down time
• 20 mins Max updates

12
How to confirm Service Levels are being
met? Normally the registry operator will perform
there own SLA monitoring and report back any
variations to the governing body. However you may
like to measure these things yourself.
13

• Service levels such as
• Overall Availability
• Per Name Server Availability
• Round Trip Time - RTT
• Packet Loss
• Average Resolution Time
• DNS updates
• Can be measured through sampling

14
Taking samples of each measurement over time
periods is the easiest and most effective way to
ensure service levels are being met. The
frequency of these samples is flexible and can be
decided between you and your registry operator.
15

• Recommended sampling Frequency
• Overall Availability ( 1 min )
• Per Name Server Availability ( 1 min )
• Round Trip Time RTT ( 1 min )
• Packet Loss ( 1 min)
• Average Resolution Time ( 1 min)
• DNS updates ( 10 min )

16
Example Using mrtg to sample Average Response
Time
Daily' Graph (5 Minute Average)
                                                  
                                        
17

• How to maintain good Service Levels?
• Choose excellent carry class data centres with
  redundant network connections, power etc
• Run multiple secondaries in diverse locations and
  networks
• Maintain a strong and up to date understanding of
  DNS
• Use clustering at each location
• Use quality hardware with different OS
• Maintain industry best practices, such as those
  stated in RFC2182, 2870 etc

18

• How to co-operate with users to maintain service
  levels?
• Target larger commercial ISPs for secondary
  services
• Making use of the existing experienced DNS
  companies
• Keep up to date, and provide discussion lists so
  that you can hear your communities needs.
• Regularly participate in discussions and show you
  are taking responsibility for the Name space

19

• In summary - key factors to maintaining SLAs are
• Multiple secondaries
• Network carrier diversity
• Provider diversity
• Monitoring
• Accountability
• Maintain industry best practice and leverage
  existing industry experience

20
Thanks for listening Any Questions?