PATHWAY Domains and ACS in ACTIVE-ACTIVE Solution

To Maintain ACTIVE-ACTIVE and main requirements for Disaster Recovery is,

 -  Maintaining a copy of the application database on remote systems.

 -  The application must be resident on the remote systems as well.

 -  Equally important is that the applications and the application environment be identical (or at least similar) on all systems in order to avoid failover faults. This requirement adds to the complexity of managing active/backup configurations and exposes these systems to human error.

HP NonStop TS/MP offers an elegant solution to this conundrum, Pathway domains. In a single NonStop system, Pathway provides an environment (PATHMON) that implements application scalability by distributing workloads across dynamic pools of application server processes (serverclasses).

A serverclass spans multiple processors within the NonStop system. The number of servers in a serverclass can be dynamically scaled up or down to meet existing transaction loads. Pathway’s Application Cluster Services (ACS) distributes transactions to servers in a serverclass to

maintain load balancing.

If a processor fails in a NonStop server, ACS distributes transactions to application servers residing on surviving processors. Thus, recovery from a processor failure in a NonStop system

is virtually instantaneous. PATHMONs are a fundamental mechanism of the fault-tolerant capabilities of NonStop servers.

PATHMONs can be distributed across multiple NonStop servers in an application network. Such a set of distributed PATHMONs is called a pathway domain. In a Pathway domain, serverclasses are distributed not only among processors with in a NONSTOP NODE but also among Nonstop nodes in an Application Network.

Pathway domains furnish 3 important advantages for business continuity.

1) The system administrator, when configuring the domains, can ensure that all processes in a server class across all domains are derived from a common executable. Consequently, The Application Processes are automatically consistent across all NONSTOP NODEs, thus eliminating failover faults.

2) If a NONSTOP System fails, further transactions are routed instantly by ACS to application processes running in surviving nodes. Thus, recovery from a system failure in instantaneous.

3) Since all of the domains in a Pathway Domain are independently configured, upgrades may be made with ZERO downtime by taking down a domain, upgrading it , returning to service and rolling the upgrade through the other domains at the same time.

To avoid a single point of failure, that portion of each serverclass resident in a NonStop server node should have its own local application database.

These databases must be kept synchronized so that the applications in each node have the same view of the application database. It is a function satisfied by the use of Shadowbase active/active replication. Shadowbase ensures that an update to any one database is immediately reflected in all of the application’s distributed databases.