Scaling and Load Balancing

Vertical Scaling refers to increasing the processing capability of a host system. This is accomplished by adding processors, memory, faster storage, and so on. Vertical scaling is primarily a hardware effort that does not affect the system topology or software configuration.

Horizontal Scaling refers to increasing the number of hardware systems hosting the software. For example, two hardware hosts running iWay achieve roughly double the throughput of one, assuming other dependent resources are available and adequately performing. Effectively distributing the workload across two or more iWay Service Manager instances is referred to as load balancing and is a key factor in achieving maximum throughput with horizontal scaling. Supporting adequate throughput is an important aspect of HA, because while a system may be online, if it is running at or close to capacity, it may appear unavailable to clients. Scaling directly addresses the throughput issue, and also provides some of the benefits of failover because it eliminates a single point of failure. Failure of one host (out of two or more) will not make the supported service(s) unavailable, although it may impact throughput and response times until the failed host has failed over to its backup or is brought back online.