Continuous availability is a design methodology that provides uninterrupted access to application services and data. Critical business applications built using this methodology are always on and available. At a minimum, the components in an application service stack are built to a 2N design specification; that is two times the Need in parallel with both sets of the Need actively being used. 2N application designs can tolerate the loss of a site, server, or infrastructure service without need for failover mechanisms or recovery operations. The approach leverages a multi-site active/active infrastructure, object brokers, and advanced storage virtualization technology to deliver a higher level of service availability than previously possible, while reducing costs and simplifying administration.
Who uses continuous availability, and why
Continuous availability architectures and processes are used by forward-thinking organizations that seek to ensure exceptionally high levels of service availability for mission- and business-critical applications. Two factors are driving this migration to continuous availability solutions. The first is the escalating demand for uninterrupted availability of key business applications, as consumers and markets have come to expect that services will be accessible anytime, from anywhere, without exception. In this context, the costs of even brief outages can be severe.
The second driver is the shortcomings of conventional approaches to the availability challenge, which is to pair separate and distinct systems and processes for high availability and disaster recovery using N+1 designs. Following conventional approaches, critical applications are architected using N+1+1 designs, that is, they are built with a Need plus one spare methodology (+1, within a single site) and disaster recovery (+1 again across primary and backup sites). And typically since the spares are passive, some sort of failover or recovery operation needs to take place to bring the spare online resulting in needless downtime.
The flaws of this approach include:
- Excessive complexity and cost, as administrators deploy and manage disparate systems, and availability topologies (HA in-site, and DR inter-site) where standby resources sit idle unless needed.
- The expectation, despite the expenditure and management effort, is that services will nevertheless suffer some costly period of downtime in the event of a disaster.
How continuous availability works
Continuous availability is made possible by technology advances that make it feasible and economical to build and maintain multi-site active/active architectures for business applications. Rather than having costly resources idling in passive mode, continuous availability topologies efficiently “stretch” active assets across two or more sites.
The main elements of a continuous availability architecture are:
- Two or more data centers deployed in active/active mode, each with a complete set of infrastructures that have no single point of failure. The sites need to be far enough apart to not both be impacted by a single disaster, but close enough to allow for synchronous data mirroring across sites. Depending on regional location, between 25 and 60 miles apart is ideal.
- An advanced distributed data virtualization layer such as EMC VPLEX Metro to synchronously mirror and maintain data coherency across sites.
- Business applications architected to run in continuous availability mode using scale-out clustering with DB solutions like Oracle RAC or clustered file systems from VMware. Oracle RAC and VMware’s VMFS file system act as object brokers to enable coherent simultaneous updates to data that is read/write in both sites.
- A transaction distribution mechanism to balance transaction loads across the active/active sites during normal operations, and to route all traffic to the good site in the event that disaster or maintenance activity takes a site down.
Benefits of continuous availability
Business applications built using continuous availability design tenets are always on and available and can tolerate the loss on an infrastructure service, application or database server, storage array, or the complete loss of a site. Administration and operations are simplified because traditional in-site HA and traditional inter-site DR solutions are combined using a common CA or continuous availability architecture. Maintenance is simplified and non-disruptive, because an entire site can be taken offline for maintenance activities. Using continuous availability designs eliminates most of the common causes of downtime and application services will have higher degrees of availability.