A scalable object location service can enable users to search for various objects in an environment where many small, networked devices are attached to objects. We investigate two hierarchical, self-configuring or unattended approaches for an eficient object location service. Each approach has its advantages and disadvantages based on the anticipated load. The first approach, SCOUT-AGG, is based on aggregation of object names. The second approach, SCOUT-MAE is based on indirection, where information about an object is stored at the locator sensor for the object. The relative efJiciency of SCOUT-AGG and SCOUT-MAP can be characterized by the query to mobilit>l update rate of the system. SCOUT-AGG performs better for low query to update rate but its performance deteriorates in general relative to SCOUT-MAP as the query to update rate increases. The rate of performance deterioration depends on query specijicity (i.e., queries for a specific object or for any object of a particular t>lpe). SCOUT-MAP generally exhibits better load balancing than SCOUT-AGG for various scenarios. We support the above results through simple analytical modeling and siniulation. 1.

Today's global internetworks challenge the ability of name services and other information services to locate data quickly. We introduce a distributed active catalog and meta-data caching for optimizing queries in this environment. Our active catalog constrains the search space for a query by returning a list of data repositories where the answer to the query is likely to be found. Meta-data caching improves performance by keeping frequently used characterizations of the search space close to the user, and eliminating active catalog communication and processing costs. When searching for query responses, our techniques contact only the small percentage of the data repositories with actual responses, resulting in search times of a few seconds. We implemented a distributed active catalog and meta-data caching in a prototype descriptive name service called "Nomenclator. " We present performance results for Nomenclator in a search space of 1000 data repositories. 1. Introduction Users canno...

ABSTRACT: Hierarchies provide scalability in large net-works and are integral to many widely-used protocols and applications. Previous approaches to constructing hierar-chies have typically either assumed static hierarchy configu-ration, or have used bottom-up construction methods. We describe how to construct hierarchies in a top-down fash-ion, and show that our method is much more efficient than bottom-up methods. We also show that top-down hierarchy construction is a better choice when administrative policy constraints are imposed on hierarchy formation. 1

We present a troubleshooting approach for coordinating problem diagnosis, and describe Global Distributed Troubleshooting (GDT), a distributed protocol which realizes this approach. We show through simulation that GDT scales well as the number of observers and problems grows.

Nomenclator is an architecture for providing efficient descriptive (attribute-based) naming in a large internet environment. As a test of the basic design, we have built a Nomenclator prototype that uses X.500 as its underlying data repository. X.500 SEARCH queries that previously took several minutes, can, in many cases, be answered in a matter of seconds. Our system improves descriptive query performance by trimming branches of the X.500 directory tree from the search. These tree-trimming techniques are part of an active catalog that constrains the search space as needed during query processing. The active catalog provides information about the data distribution (meta-data) to constrain query processing on demand. Nomenclator caches both data (responses to queries) and meta-data (data distribution information, tree-trimming techniques, data access techniques) to speed future queries. Nomenclator relieves users of the need to understand the structure of the name space to locate objec...

This thesis addresses the challenge of locating people, resources, and other objects in the global Internet. As the Internet grows beyond a million hosts in tens of thousands of organizations, it is increasingly difficult to locate any particular object. Hierarchical name services are frustrating, because users must guess the unique names for objects or navigate the name space to find information. Descriptive (i.e. relational) name services offer the promise of simple resource location through a non-procedural query language. Users locate resources by describing resource attributes. This thesis makes the promise of descriptive name services real by providing fast query processing in large internets. The key to speed in descriptive query processing is constraining the search space using two new techniques, called an active catalog and meta-data caching. The active catalog constrains the search space for a query by returning a list of data repositories where the answer to the query is li...

There is a proliferation of application-level servers that offer enhanced network services beyond the simple packet forwarding provided by the underlying network infrastructure. Examples of such servers range from Web server mirrors, Web caches, Web page distilling proxies, video transcoders, and application-level multicast routers, to application-level load-adaptive multipath routers. A fundamental question arising from the deployment of such servers is where to place them within a network. This paper explores technical issues related to the creation of an infrastructure to allow self-organization of network service placement based on observed demand for each service. In so doing, we propose a framework, called Sortie, whereby services are allocated on network nodes based on a set of very simple rules independently executed by each node. The distributed nature of allocation decisions ensures the scalability of the framework. We also present simulation results confirming the stability ...

Abstract-This paper presents a mechanism that facilitates and enhances the use of independently administered remote network servers in the presence of server interface heterogeneity. The mechanism is designed under the client-service model, which extends the client-server model with an abstraction of service to decouple abstract server capabilities from concrete server inter-face specifics such as server interface binding protocols and the interface operation invocation protocols. The mechanism selects servers, accommodates server interface heterogeneity, and han-dles server access failures as per the abstract server capabilities desired by the client. It could return the identity of the server used for each service access invocation to facilitate billing, refining service specifications, and reporting server-specific errors. This paper also illustrates a C library interface to this mecha-nism, and describes a language veneer over the C programming language demonstrating how a typed procedural language could be extended by a few language constructs to support the mech-anism under the client-service model. In this language, server capabilities are referenced by abstract data type (ADT) objects, and are accessed by invoking the objects ’ interface operations using a call-by-value-result paradigm. This language veneer also makes it easier to port the client software across to systems that use different service specification schemes. Our work suggests that this mechanism facilitates the develop-ment, use, and maintenance of client and server software in large heterogeneous distributed systems comprising many autonomous servers. It also shows that the overhead of invoking remote server operations via the mechanism can be quite low. Index Terms-Heterogeneous distributed systems, client-server model, client-service model, service acquisition mechanism, attribute-based naming, remote procedure call, server interface directory service, agent process, fault tolerance, object-oriented programming, language veneer. I.

This paper explores an attribute-based approach to storing information in the context of a #le system that supports extended attributes about #les and a mechanism to manipulate #les based on logical queries and comparisons of attributes. The novel aspects of our system are that it is sophisticated enough to operate as a user's primary method of interaction with the operating system and that it supports derived attributes whose values are derived at look-up time. Organizational mechanisms are explored to aid in the navigation of the system. We implemented our system as a user-level NFS server that supports attribute-based naming of #les and other objects. We discuss its details, our experiences with it and performance comparisons between it and a traditional hierarchical #le system. *Current address: Silicon Graphics, Inc. Mountain View, CA. 1 1 Introduction As an increasing amount of information reaches our computers the problem of managing information becomes more import...

An easily accessible name service is an important asset for the management of a network, and a standard service like X.500 could enhance considerably the operation of heterogeneous networks based on the OSI suite. However, the naming scheme used by X.500 is complex, and not necessarily user friendly. In this paper, after a short presentation of X.500 operations and names, we will justify a proposition to use untyped "short forms", made of sequences of text tokens, and we will develop a set of algorithms for accessing the directory with these short names, before giving comparison and measures of these algorithms. This short form of names is compatible with the naming schema used in Internet, so that we could retain the simplicity and ease of use of the Internet naming system, while being compatible with the complete X.500 standard. Our solution is particularly useful for the category of names that can be found for example when interfacing mail to X.500 and where resolution tim...