Skip to end of metadata
Go to start of metadata

Service Level Agreement

We are proud to offer our customers a service level guarantee backed by a 24/7 network operations staff.

Every AppNexus service is designed for fault tolerance and redundancy.  However, servers can, will, and do crash.  It is the client's responsibility to use load balancing and other techniques to ensure that the failure of any particular server will not impact end-user experience or result in data loss.

Monitoring and Unavailability

Servers

The API will be monitored in each datacenter by a script executing a simple command every 5 minutes.  If this fails, other than during a maintenance window or scheduled outage, the Service will be considered unavailable.

Local Load Balancing

Every five minutes, AppNexus will ping each load balancer.  If at least one of each pair of load balancers does not respond to a ping, load balancing will be considered unavailable.

Global Load Balancing

AppNexus will use a third-party monitoring service (Gomez, Keynote, AlertSite, or equivalent) to track availability of the Global Load Balancing Service from at least 5 locations around the United States and Europe by making a DNS lookup request.  If the DNS request fails, or takes more than 5 seconds, the Global Load Balancing Service will be considered unavailable.

Content Delivery (CDN)

AppNexus will use a third-party monitoring service (Gomez, Keynote, AlertSite, or equivalent) to track availability of the CDN from at least 5 locations around the United States and Europe.  If a CDN request fails, or takes more than 5 seconds, the CDN service will be considered unavailable.

Storage

Every five minutes, AppNexus will attempt to access the client's file system.  If standard file operations fail, the storage service will be considered unavailable.

Maintenance Windows

From time to time, AppNexus may need to perform maintenance on one or more services.  Except in the case of emergency, AppNexus will provide at least two business days notice before performing maintenance.  If maintenance will cause a service interruption in a particular datacenter, AppNexus will provide a clearly documented procedure for client to avoid any end-user disruptions, and will provide at least ten business days notice.