High Availability DNS and DHCP with the AdonisDNS Crossover High Availability (XHA™)Implementing a properly configured, highly available, and distributed DNS solution costs far less than having an outage (even for half an hour). Even in multiple-server DNS configurations, the failure of a primary server can cause serious problems. If the data on the secondary servers expires before a new primary becomes available, all the secondary servers will cease to answer queries. Updates cannot be made to the DNS configuration on the secondary servers without a primary server. To counter this, Adonis appliances can be connected into a high availability pair, forming a clustered primary server. The two combined servers handle fail-over conditions, with individual nodes operating in an active/passive mode. During a failure, the passive node will take control in less than two seconds, and will handle the traffic load. DNS was designed to be a scalable, distributed service able to handle failure and partial outages. Multiple authoritative servers per domain allows for a server to fail while another operating server responds to a request. In such a deployment, many organizations use multiple servers on different networks, ensuring that at least one server will respond. The drawback is that a failed response increases the latency for DNS resolution, sometimes from a few milliseconds to several seconds. Such delay makes applications seem sluggish, and users often perceive poor DNS performance as poor application or system responsiveness, sometimes causing DNS resolution to fail completely. Many organizations use an architecture based on a primary-secondary relationship where the secondary servers are authoritative for data as long as the primary’s grant authority. Each secondary server is given an expiry time, after which it becomes non-authoritative and enters into a failure state. Even with many DNS servers, a request can be denied if the primary server goes offline and the secondary servers expire. The solution to such issues is to create a high availability (HA) cluster for DNS services. With an HA cluster, DNS queries can be resolved even in the event that one of the DNS servers fails. For internal networks, this reduces resolution latency, and the volume of help desk tickets. The Adonis HA solution uses two servers in an active-passive cluster. The two servers share an address that clients query. The servers connect over the IP network to determine the health of the active unit. In the event that the active unit fails, control is transferred to the passive server that takes the active role and the shared address. However, BIND was not designed with HA in mind, and many existing HA systems for DNS handle dynamic updates poorly. The problem: most HA configurations are centered on a primary server and the active node will receive the update, but the passive node will not be notified. Crossover High Availability (XHA™) uses an enslaved primary as the passive node. When updates are sent to the active node, updates are automatically propagated to the passive node as standard incremental zone transfers. 
Adonis XHA™ ArchitectureIn addition to XHA™ as a comprehensive high availability solution for redundancy, Adonis also introduces an HA “self-healing” system. The self-healing aspect enforces that a valid configuration exist before a passive server becomes part of the cluster. This system guarantees that a passive server is always synchronized with its active partner, even if the passive server is unavailable during configuration deployment. A high availability repair system that resolves broken HA configurations and eases hardware swap-outs is also included. DHCP Crossover High Availability (XHA™)Adonis securely alleviates traditional drawbacks to redundancy with failover, resiliency, and disaster recovery without splitting IP addresses. The revolutionary Crossover High Availability (XHA™) solution uses an enslaved primary as the passive node, which is automatically propagated with updates sent to the active node. If the active server fails, the passive server becomes the new active server, readily taking over the role of issuing IP addresses from the same address pool. When the original active server is restored, it becomes the new passive server, which is then constantly updated by the new active server. This design ensures seamless failover without downtime or manual intervention, fully optimizes the utility of scarce IP addresses, and guarantees that a passive server remains synchronized with its active partner. 
|
