MESIS has been designed with the goal of providing an integrated environment that addresses all typical management issues of complex organizations. Organizations usually consist of several departments, even geographically distributed over the Internet. Each department has its private LAN, and needs to interact via gateways with other departments to accomplish coordinated tasks. The provision of an Internet service, for instance, may span several organization boundaries. In addition, when different departments of one organization have to communicate via the Internet, the system should grant the same levels of security and QoS of intranet communication. Moreover, the globality of the scenario addressed by MESIS imposes to face up to the issue of scalability.
For that reason, we have considered fundamental to introduce and to make possible the handling of the locality concept: MESIS embeds locality via a hierarchy of locality abstractions suitable for describing global distributed systems, ranging from simple LANs to the Internet (see Figure 1). Any node part of the MESIS environment hosts at least one place for agent execution and management; several places are grouped into domain abstractions. In each domain, a default place hosts a gateway which is in charge of inter-domain routing functionality. The locality permits also to introduce a scope when considering all other system policies, and helps in granting a protected framework for the entities that belong to the locality itself.

The core part of the MESIS project is its architecture that provides a distributed infrastructure with a set of facilities for the design and the development of complex network-centric applications (see Figure 2). All facilities are implemented on top of the MESIS DPE layer by a set of coordinated mobile agents.
In addition, the openness property of the MESIS infrastructure permits to extend the programming framework by dynamically adding new services, even built on the already provided functionality.
MESIS DPE facilities are split in two levels, the lower one that groups the basic and primary mechanisms, the upper one that comprehends more evolved tools and services. The upper facilities generally use the lower ones in their implementation; for instance, the naming facility, that keeps track of entities location, makes use of the underlying identification facility.