Reliability. Unlike a server farm (an independent set of servers), clustering provides the reliability required for mission-critical applications. Distributed applications must address network, hardware, and software errors that might produce duplicate (or failed) transactions. Clustering makes it possible to deliver "exactly once" execution as well as checkpoint/restart functionality for critical operations.

Availability. Clustering provides virtually uninterrupted availability by deploying applications on multiple Integration Servers; in the worst case, a server failure produces degraded but not disrupted service. Without clustering-even with expensive Fault-Tolerant systems-a failure of the system (hardware, java runtime, or software) may result in unacceptable downtime.

Scalability. Clustering provides horizontal scalability, which allows virtually limitless expansion of capacity by simply adding more machines of the same or similar capacity. Without clustering, only vertical scalability is possible. That is, increased capacity requirements can only be met by deploying on larger, more powerful machines, typically housing multiple CPUs. Salesforce.com enforces a usage metering for calls via its API, in order to balance transaction loads. This is done by limiting the number of API requests (or calls) that any one organization may execute concurrently or within any 24-hour period. These limits are based on the type of edition subscribed by an organization and are against the aggregate of all API calls, and not on a per user basis. When an organization exceeds the limit, the users may be temporarily blocked from making additional calls.