Discovering Dependencies for Network Management

It is lamentable that Leslie Lamport’s famous quote [9] “A distributed system is one in which the failure of a computer you didn’t even know existed can render your own computer unusable” describes a scenario familiar to almost every computer user. As IT systems are increasingly distributed, it is not only the clients and servers themselves that can render a computer useless for an afternoon, but any of the many routers, links and network services also involved. In distributed systems, the underlying problem is the absence of tools to identify the components that “can render your own computer unusable”: the implicit web of dependencies among these components exists only in the minds of the human experts running them. The complexity of these dependencies quickly adds up, requiring more help than traditional IT management software provides. Listing the contents of a single DFS 1 directory, for example, can involve a minimum of three hosts and eight network services (WINS, ICMP Echo, SMB, DFS, DNS, Kerberos, ISA key exchange, ARP). Existing management solutions focus on network elements, topology discovery, or particular services, but what is needed are tools to manage and improve the user’s end-to-end experience of networked applications. In deference to Lamport, this paper defines the Leslie Graph as the graph representing the dependencies between the system components, with subgraphs representing the dependencies pertaining to a particular application or activity. Nodes represent the computers, routers and services on which user activities rely, and directed edges capture their inter-dependencies. Different versions of a Leslie Graph can express different granularities of dependence for an activity — for some analyses, an Leslie Graph capturing intermachine dependences at the granularity of IP addresses might be sufficient, while for others an Leslie Graph capturing inter-service dependencies at the granularity of software processes might be desirable. This paper makes three contributions: (i) we define Leslie Graphs and discuss the challenges in finding them, (ii) we suggest important problems that Leslie Graphs could help