Like dedicated hosting, cloud computing has to address availability. Continued cloud outages, and the corresponding publicity, remind us of the importance of resiliency and availability. One of the major benefits of cloud computing is scalability and efficiency of multi-tenant infrastructure. However, even cloud infrastructures have to run in a physical data center somewhere, bringing us back to the critical nature of infrastructure availability.
Fortunately, the same availability you are accustomed to as part of a dedicated environment can be found in cloud computing. Availability can be viewed in a continuum that ranges from high availability to failover and recovery, with many nuances in-between. This continuum of availability enables clouds to fulfill enterprise application and business needs at many different price points.
Platform Resiliency for Continuous Uptime
The first area to address availability is the resiliency of the platform itself. Businesses requiring enterprise-class infrastructure need to look under the hood to determine how the infrastructure is architected and how resiliency is addressed. A highly resilient environment should automatically
detect and address the failure of a system component—whether it is a server, network, a full blade or the VM —to quickly shift to a redundant component in order to keep the application running in the current site.
Failover is the capability to switch to a redundant or standby computer server, system, or network upon the failure or interruption of the primary environment. Cloud computing has allowed failover practices to become less reliant on physical hardware and therefore more
available and less costly. Service providers vary in the type of fail-over they provide as well as the time to respond, depending on the customers’ RPO and RTO needs.
A failover, or warm failover can be used for applications that require slightly less than real-time (e.g. hours VS. seconds). In warm failover, a second site stands ready to be activated and made current as quickly as required. Depending on the need, the time to failover depends on the Customer’s recovery time objective. Sometimes the options can include the secondary site begin brought on line using a previous copy of the primary site. Usually the copy is from the previous day, but it can be older depending on the business need.
High Availability for Mission-critical Apps
High availability addresses mission-critical production systems that require immediate, continuous, 24/7 access to data. More technically, it means data must be duplicated at another location, usually in a different geographic area. Essentially you are renting resources at one location and identical resources at another location, so costs are higher.
The communication method used between the systems also affects availability and costs. Synchronous near real-time communication pdates data from the primary system immediately to the secondary system. The secondary system mirrors the first and is ready to go into operation if the first system fails for any reason.
Asynchronous communications is where data waits in queue until the second system is free to accept it, so by its nature is less real-time. Again, the business need determines which communications method is better.
Recovery for Availability
Recovery represents the other end of the availability continuum. Cloud computing is changing the disaster recovery landscape. The scalability and
flexibility of cloud computing platforms enable higher application availability. Recovery can be used as a back-up to a production system already in the cloud or as a recovery solution to another data center. Further, the back-up can be on-line, ready to operate at the cloud site (like a warm failover) or off-line at a cloud site, as done in traditional recovery scenarios, since the cloud is a cost-effective recovery site for legacy systems.
As is obvious, different applications require different levels of availability, and applications should not be shoehorned into a “one size fits all” cloud
environment. The best cloud providers will work closely with you to understand the business requirements of your business applications and devise the appropriate level of availability for each application you want to move to the cloud, along with any need for cloud resources to facilitate recovery of applications you do not move to the cloud.