Domain name registries provide some of the most essential functions of the modern Internet. At the same time, the Internet is increasingly relied upon to perform critical functions for companies, organizations, and individuals throughout the world. These realities dictate that the reliability of a new registry’s functions must approach 100% despite the constant need for upgrades to both hardware and software. The Registry’s technical team has taken these demands into consideration and designed systems with reliability that is unparalleled on the Internet today.

RegisterOrg commits to quality of service levels that meet or exceed the best-of-breed SLA’s of any existing gTLD. The Registry will achieve DNS updates within 5 minutes, DNS uptime of 100%, and DNS responses of <300ms, Whois responses of <800ms and SRS checks of <400ms.

RegisterOrg has analyzed the Service Level Agreements in place for the existing gTLDs to determine the best-of-breed service levels for DNS, Whois and SRS performance and reliability. The following table summarizes the SLAs in place for .name, .info, .biz, .pro and the existing .com/.net/.org registry run by VeriSign to discern the best-of-breed levels.

	.name	.info	.biz	CNO (VRSN)	.pro	Proposed .org
DNS uptime	99.999%	99.930%	99.999%	N/A	100.000%	100.000%
DNS response time (secs)	N/A	0.3	1.5	N/A	0.3	0.3
DNS update time (minutes)	5	5	15	N/A	15	5
Whois uptime	99.40%	99.79%	99.95%	N/A	99.79%	99.99%
Whois response time (secs)	N/A	0.8	1.5	N/A	1.5	0.8
Whois update (minutes)	"near real time"	N/A	15	N/A	15	15
SRS uptime	99.40%	99.93%	99.90%	99.44%	99.87%	99.99%
SRS check response (secs)	3	0.4	1.5	3	1.5	0.4
SRS write response (secs)	3	0.8	3	5	3	0.8

Due to the importance placed on the reliability and stability of the .org registry, RegisterOrg commits to service levels that meet or exceed the best-of-breed SLAs for the existing gTLDs, as shown above. As described in C17.10, the Registry’s sub-contractor has the systems in place to meet or exceed these SLA guarantees.

In addition to setting the highest hurdle rates for service levels in the industry, RegisterOrg has also sought to define the strictest service metrics in the industry, as described in C17.13.

The remainder of this section will define each of the service metrics. In addition to setting the highest hurdle rates for service levels in the industry, the Registry will also seek to define the strictest service metrics in the industry.

The first step in any definition of system reliability must involve some attempt to define what it means to be “up” versus “down”. This is essential because a service can be running, but with performance so degraded as to be useless. We submit that in order for registry services to be useful to the Internet community, they must not only be available, but responsive. Issues relating to performance must be an essential part of the definition of service reliability, and RegisterOrg has used this as a basis to create the following quality-of-service definitions:

Service description	Uptime service definition
Shared registry service (RRP)	The shared registry service shall be considered to be up during any minute-long interval in which 95% of add, modify, and delete transactions are successfully completed within 800 milliseconds, and 95% of check transactions are completed within 400 milliseconds.
Whois	The Whois service shall be considered to be up during any minute-long interval in which 95% of all requests return correct data within 800 milliseconds.
DNS (cluster)	An individual DNS cluster shall be considered to be up during any minute-long interval in which 95% of all requests to the cluster return a correct response within 300 milliseconds.
DNS (service)	The DNS service shall be considered to be up during any interval in which at least half of the total clusters are considered to be up.

The quality of service definitions above assume that the measurement is being made from the same network as the measured service. In order to measure the responsiveness of .org registry services from various portions of the Internet, the Registry will contract with a service measurement company such as Keynote or Service Metrics to provide data regarding the performance of core .org registry functions.

The Registry’s system will provide DNS service for the .org TLD that end-users throughout the Internet will rely on extensively, and without which routine Internet functions relating to .org domains will not be possible. The Registry’s provider will devote significant resources to ensure that its DNS service is extremely reliable (see C17.4, C17.5, and attachment, Section XI, DNS). By employing redundancy at every level of its design, including the deployment of a parallel DNS infrastructure running BIND, Register.com is able to commit to a 100% SLA uptime for the DNS service. The uptime of any individual DNS cluster will be guaranteed to be at least 99.95% during any month-long interval.

The Registry’s global DNS constellation is designed to minimize network latency regardless of the origin of the query. In order to measure the cross-network performance of the DNS constellation for .org, ICANN has proposed the CNNP tests described in Appendix D of the model registry agreement. RegisterOrg is fully prepared to meet ICANN’s CNNP testing requirements.

The .org Whois and shared registry services are other important public resources, and the Registry’s systems will be engineered to provide significant redundancy and ensure high availability. The advanced design of these systems allows Register.com to guarantee a 99.95% uptime for both Whois and shared registry services.

In order to perform periodic maintenance, Registry’s technical team will also schedule maintenance windows for two hours each week. Although this weekly maintenance window will exist, the Registry recognizes that any downtime can be extremely disruptive to both registrars and end-users, so maintenance windows will only be used under the following circumstances: Registrars will be notified at least 72 hours prior to the maintenance window; and No more than 10 maintenance windows will be used during any year.

Shared registry service or Whois downtime during maintenance windows will not be considered in evaluating the uptime guarantees described earlier in this section. Due to its critical nature, there will be no scheduled downtime windows for DNS service.

Additionally, for periodic major upgrades to the system, the Registry will be permitted to include up to two extended outages per year. These outages may be up to six hours in duration and must be announced to registrars at least 15 days in advance.