The Unified Modeling Language User Guide SECOND EDITION [Electronic resources] نسخه متنی

Getting Started

When you start to model most software systems, you can usually assume a frictionless environmentmessages are sent in zero time, networks never go down, workstations never fail, the load across your network is always evenly balanced. Unfortunately, the real world does not work that waymessages do take time to deliver (and, sometimes, never get delivered), networks do go down, workstations do fail, and a network's load is often unbalanced. Therefore, when you encounter systems that must operate in the real world, you have to take into account the issues of time and space.

A real-time system is one in which certain behavior must be carried out at a precise absolute or relative time and within a predictable, often constrained, duration. At one extreme, such systems may be hard real time and require complete and repeatable behavior within nanoseconds or milliseconds. At the other extreme, models may be near real time and also require predictable behavior, but on the order of seconds or longer.

A distributed system is one in which components may be physically distributed across nodes. These nodes may represent different processors physically located in the same box, or they may even represent computers that are located half a world away from one another.

To represent the modeling needs of real time and distributed systems, the UML provides a graphic representation for timing marks, time expressions, timing constraints, and location, as Figure 24-1 shows.