Flyover: Network Management

        Perspectives
        Services
        Dimensions
        Tools
        Concepts, Themes and Wisdom

What Goes Wrong with Networks

        Growth
        People and Processes
        New Applications
        Traffic Shifts
        Equipment Fails

Layers of Network Management

        end-user
        lan manager
        can manager
        enterprise-wide
        ISP

Points of View

        End-user - "make it do what I want, make it seamless"
        CAN manager - "keep it working, make it scale and stay in budget"
        ISP - "stay ahead of the train, and coopetition with peers"

End-user Management Goals

        Reliability
        No barriers
        Consistent interface
        Reasonable response time

User Expectations

        Resolving network problems and providing second and third level support
        Providing one-start shopping and closure of problems
        Everything behind the wall plate
        Implementation of moves and changes, network expansion and
reconfiguration
        Network design, network modeling and capacity planning
        Client billing and accounting
        Security management
        Management of change

CAN Management Goals

        Cost-effective performance
        Robustness
        Multistack desktops and lan integration
        Security
        Enterprise-wide networked computing environment
        Working with the end-user
        Working with their ISP

ISP Management Goals

        Figure out what's happening
        Cope with growth
        Maneuver with other ISP's
        Minimize packet loss
        Stay within budget

Management Services

        Standard administrative services
        Fault management
        Configuration management
        Accounting management
        Security management
        Network and server performance monitoring
        Application management

Standard local services

          Create and control accounts for hosts and for nets
          Tracking cables and connections
          Print services
                slave             -net  (single and multiprotocol)
         File services and disk farms
        Backups - server and clients
        Archives
        Access controls and security
        Gateway connection
        Problem reporting
        Software distribution, license servers, mail configuration  and
change management
        Middleware (sqlnet, rpc runtimes, etc)
        Underware (name service, time service, etc.)

Enterprise-wide Services

        Network Management Central
        NMS   - Organization   - Central Layer 3+
        Desktop Support
        Enterprise-wide Identity and Directories
        Enterprise-wide Application Support
        Internet Services
        Security   - Web    - ISP Relations
        ISP Management Issues
        Accounting

Dimensions of Management

        technical
        financial
        political

Technical

        topology
        hardware selection
        protocols - which ones, which layers
        optimizations (performance, manageability, reliability, redundancy,
diagnosability, expansion and upgrade strategies)
        reengineering

Financial

        build or buy
        expenses and revenue
        centralized and distributed costs
        charging and cost-recovery

Political

        control and responsibility
        policies and procedures
        formal and informal authority
        levels and loci of risk

Tools

        Physical Tools - tdr, sniffers, routers, switches, fault management
        Logical tools  - SNMP, RMON, firewalls, traceroute, performance
monitoring
        Internet tools -subnets, routing, dns, security
        Staff tools - NICS and NOCs
        Process Tools - service centers, trouble tickets, training  NMS

Fundamental Concepts

        layering and integration
        monitoring
                statistics
                performance
        relationships - peering, hierarchy, heterarchy
        symmetry and asymmetry
        Interoperability  - at technical and management layers
        Cost - capitalization and operating
        Error handling
        Upward migrations and embedded bases

Recurring Themes

        Standards
        Tradeoffs
                space vs time
                delay and throughput
        Measuring performance
        Security
        Reengineering
        Configuration
        Trouble tickets and process
        Central versus distributed approaches
        Collegial and competitive relationships

Wisdom

        One standard and many
        A bad application can ruin a good network
        Manage change
        Manage expectations
        Management of technology  is always an afterthought
        Speed of light is a constant
        Clue is a constant
        Suboptimal designing makes sense
        Security: the right tool for the right job
        Technical solutions to political problems frequently work
        Technical solutions to economic problems seldom work