Who manages who?

The question of whether system management is part of network management or vice versa will be hotly debated in the coming year. Our view is that the network extends from our finger tip to the application executable code, so that even queue management and caching control inside the servers are right and proper network management functions. There are stumbling blocks and expenses that inhibit this complete view of network management. As client/server systems get deployed, savings are not as great as expected because they are much more expensive to operate and evolve than expected. The "administrator to box" ratio remains constant because of the poor systems and network management design which is slowing down progress. Planners are not faced with a "buy or build" decision since no solution solves the existing problems, and none can be built fast enough to keep up with the demand. Administrators are often faced with the problem of integrating solutions from different suppliers. People are beginning to use the internet for remote management to good effect, but this is a "send and pray" network that cannot be counted on under stress to isolate and fix problems. It may prove to be perfectly fine for detecting problems, but will it be adequate to force test messages on defined paths, or to monitor major network events, or to isolate focused overloads, or to convey commands to network elements? A special challenge is managing the hardware and software for PC desktops and servers. The Desktop Management Task Force is creating a set of specifications which can break the barrier to widespread adoption of client/ server computing. They address the issues of consistent software upgrades to the clients across different networks. System Management technology will grow once a standard for a 'Desktop Management Interface' is adopted and built into products. Until then, managers will use a battery of separate vendor-developed utilities, many of which cannot run remotely, to manage their client/server networks.