Directly related to the Base architecture is the TACC [11] platform, which provides a cluster-based environment for scalable Internet services. In TACC, service components (or ``workers'') can be written in a number of languages and are controlled by a front-end machine which dispatches incoming requests to back-end cluster machines, incorporating load-balancing and restart in the case of node failure. TACC workers may be chained across the cluster for composable tasks. TACC was designed to support Internet services which perform data transformation and aggregation tasks. Base services can additionally implement long-lived and persistent services; the result is that the Ninja approach addresses a wider set of potential applications and system-support issues. Furthermore, Base services can dynamically created and destroyed through the iSpace loader interface on each MultiSpace node--TACC did not have this functionality.
Sun's JINI [31] architecture is similar to the the Base architecture in that it proposes to develop a Java-based lingua franca for binding users, devices, and services together in an intelligent, programmable Internet-wide infrastructure. JINI's use of RMI as the basic communication substrate and use of code mobility for distributing service and device interfaces has a great deal of rapport with our approach. However, we believe that we are addressing problems which JINI does not directly solve: providing a hardware and software environment supporting scalable, fault-tolerant services is not within the JINI problem domain, nor is the use of dynamically-generated code components to act as interfaces into services. However, JINI has touched on issues such as service discovery and naming which have not yet been dealt with by the Base architecture; likewise, JINI's use of JavaSpaces and ``leases'' as a persistent storage model may interact well with the Base service model.
ANTS [30, 29] is a system that enables the dynamic deployment of mobile code which implements network protocols within Active Routers. Coded in Java, and utilizing techniques similar to those in the iSpace environment, ANTS has a similar set of goals in mind as the Base architecture. However, ANTS uses mobile code for processing each packet passing through a router; Base service components are executed on the granularity of an RMI call. Liquid Software [19] and Joust [15] are similar to ANTS in that they propose an environment which uses mobile code to customize nodes in the network for communications oriented tasks. These systems focused on adapting systems at the level of network protocol code, while the Base architecture uses code mobility for distribution of service components both internally to and externally from a Base.
SanFrancisco [7] is a platform for building distributed, object-oriented business applications. Its primary goal is to simplify the development of these applications by providing developers a set of industrial-strength ``Foundation'' objects which implement common functionality. As such, SanFrancisco is very similar to Sun's Enterprise Java Beans in that it provides a framework for constructing applications using reusable components, with SanFrancisco providing a number of generic components to start with. MultiSpace addresses a different set of goals than Enterprise Java Beans and SanFrancisco in that it defines a flexible runtime environment for services, and MultiSpace intends to provide scalability and fault-tolerance by leveraging the flexibility of a component architecture. MultiSpace services could be built using the EJB or SanFrancisco model (extended to expose the MultiSpace functionality), but these issues appear to be orthogonal.
The Distributed Computing Environment (DCE) [20] is a software suite that provides a middleware platform that operates on many operating systems and environments. DCE abstracts away many OS and network services (such as threads, security, a directory service, and RPC) and therefore allows programmers to implement DCE middleware independent of the vagaries of particular operating systems. DCE is rich, robust, but notoriously heavyweight, and its focus is on providing interoperable, wide-area middleware. MultiSpace is far less mature, but focuses instead on providing a platform for rapidly adaptable services that are housed within a single administrative domain (the Base).