Availability Requirements

The Availability requirements define our availability targets, downtime handling and reporting, scheduled maintenance windows, and other relevant requirements.

Definitions

Available
We define "Available" as the Web site being online and accepting requests. This requires that the application server, cache server, database server, and job server are all running. This requires the network connection between all the components to be functional. This requires the network connection between the application server's access point and the wider Internet to be functional.
Degraded Availability
We define "Degraded Availability" to be an instance where the system is available for processing requests, but with a violation of performance requirements or characteristics. For instance, if a network issue was causing requests to process much slower than usual, but requests were still being processed otherwise normally, this would be Degraded Availability.

Goals

  1. To facilitate commerce by being as highly available as possible within the constraints of our present resources.
  2. To communicate both planned and unplanned instances of downtime effectively to all concerned parties.

Exclusions

  1. The system handling WTI email is not included in MoR availability planning. If the email system is down, but all components of MoR are running, the system shall still be considered available.

Requirements

  1. The system shall be available for at least 99.9% of each month.
  2. The system shall have at least degraded availability for 99.99% of each month.
  3. The system may be down for maintenance on Wednesday evenings, 18:00 through 20:00 inclusive, in the prevailing time zone controlling the territory of Oklahoma (typically "Central Time"). This maintenance window shall not affect the calculation of system availability. If the system is unavailable during this window, the availability percentage shall not be reduced.
  4. The system shall be monitored by an external service for availability.
  5. Any instance of planned maintenance, degraded availability, or general unavailability shall be announced in a centralised location which is accessible to the public.
    1. The announcement must include an estimated time of resolution.
    2. When the system returns to full availability, a further announcement shall be posted noting the system has been restored.
  6. The system shall provide an option to store owners to protect from AI bot crawlers using the Anubis system. This requirement is of low priority for the initial release of MoR.