Network Topology Simulator

How to use the topology simulator

Start by adding devices from the left toolbar — click Router, Switch, Host, Firewall, or Cloud to place nodes on the canvas. Drag nodes to reposition them. Double-click a node to rename it inline.

To connect two devices, click Connect in the toolbar, then click the source device followed by the target device. The simulator automatically selects the first available interface on each end. Click a link to inspect or remove it.

Once your topology is laid out, use Auto-assign IPs to allocate point-to-point /30 subnets from the 10.0.0.0/8 pool — one per link, starting from 10.0.0.0/30. Use Analyze to generate the adjacency table, detect loops, and identify disconnected devices. Use Export Cisco Config to download ready-to-paste IOS commands for every router.

Network topology types

Choosing the right topology is the first design decision in any network project. Each type offers different trade-offs in cost, redundancy, and scalability.

Star topology

All end devices connect to a central switch or hub. The star is the most common LAN topology because failures are isolated — a broken cable only disconnects one host. The central switch is a single point of failure, which is why enterprise designs stack or dual-home the core switch. In the simulator, place a switch at the center and connect multiple hosts to it to model a star.

Mesh topology

Every node connects directly to every other node (full mesh) or to multiple peers (partial mesh). Full mesh provides maximum redundancy but the number of links grows as n(n−1)/2, making it expensive at scale. Partial mesh is common in WAN designs where critical sites get redundant paths and branch offices get a single uplink. The simulator's loop detector will flag a full mesh — use Spanning Tree Protocol (STP) or routing protocols to manage redundancy.

Ring topology

Devices form a closed loop where each node connects to exactly two neighbors. Token Ring and SONET/SDH use ring architectures. A single link failure leaves the ring intact (traffic reroutes); a second failure isolates a segment. The simulator's Analyze button detects ring loops and warns about them — useful when verifying L2 designs before enabling STP.

Bus topology

All devices share a single backbone cable with terminators at each end. Bus was standard in 10BASE-2 (Thinnet) Ethernet. A break anywhere on the bus brings down the whole segment, so it has no place in modern designs. It is modeled in the simulator by using a single switch with all hosts attached — conceptually equivalent for modern Ethernet.

Hybrid topology

Real enterprise networks combine multiple topologies. A common pattern: a redundant core mesh between distribution-layer switches, star spokes extending to access-layer switches, and point-to-point WAN links to branch routers. The simulator supports all of these simultaneously — mix routers, switches, and hosts in any combination.

Point-to-point links and /30 subnets

Router-to-router links are always point-to-point: two endpoints, four addresses. A /30 subnet (255.255.255.252) provides exactly two usable host addresses — one for each router interface — wasting no address space. The auto-assign algorithm allocates contiguous /30 blocks from the parent pool: 10.0.0.0/30, 10.0.0.4/30, 10.0.0.8/30, and so on. RFC 3021 also permits /31 subnets for P2P links, saving one more address per link.

Reading the adjacency table

The adjacency table produced by Analyze lists every neighbor relationship in the topology. Each link appears twice — once from each endpoint's perspective — matching the format used by protocols like OSPF and IS-IS in their link-state databases. The table shows the local interface, the neighbor device and its interface, and the subnet assigned to that link. Use it to verify connectivity before configuring routing.

Exporting to Cisco IOS

The Export Cisco Config button generates a configuration block for each router in the topology. Each block includes hostname, one interface stanza per connected interface with its ip address and no shutdown, and a comment header. Paste the output directly into a Cisco router's privileged EXEC mode or into a GNS3/Packet Tracer startup config.

How to design a scalable network topology

Scalable network design follows the three-layer hierarchical model: access, distribution, and core. The access layer connects end devices — workstations, phones, printers, and servers — to the network. Access switches are configured with VLANs, port security, and spanning tree PortFast. The distribution layer aggregates access switches, enforces inter-VLAN routing policies, and connects to the core. Distribution-layer switches run Layer 3, hold the default gateways for each access VLAN, and apply QoS and security policies. The core layer provides fast, reliable transport between distribution blocks and to the WAN edge. Core switches carry no user traffic directly — they forward packets as quickly as possible between distribution switches.

This separation of concerns makes the network scalable: adding a new floor means adding access switches and connecting them to the existing distribution block without redesigning the core. The distribution layer can be expanded independently. The core remains unchanged as the network grows at the edges. In smaller networks (under 200 devices), the distribution and core layers are often collapsed into a single tier — called a collapsed core — where two high-end switches serve both distribution and core functions with a redundant link between them.

Redundancy is built into every layer. At the access layer, dual uplinks from each access switch to two distribution switches protect against an uplink failure. Spanning Tree Protocol (or its rapid variant, RSTP) blocks one uplink to prevent loops while keeping it ready for failover. At the distribution and core layers, redundant links between switches are handled by EtherChannel (aggregating multiple physical links into one logical link) or by OSPF/EIGRP routing over multiple paths, which provides per-packet or per-flow load balancing in addition to fast failover.

IP addressing best practices for network design

A good IP addressing plan is hierarchical, summarizable, and documented. Start with the largest block you will ever need and subdivide it systematically. Assign address ranges by function and security zone, not by physical location alone. Users, servers, management devices, and WAN links should each occupy distinct, non-overlapping ranges so that ACLs and firewall rules can be expressed with simple wildcard masks. For example: 10.0.0.0/8 as the master block, 10.1.0.0/16 for headquarters, 10.2.0.0/16 for site A, 10.3.0.0/16 for site B, and within each /16, divide 0.0–127.x for users, 128.x–191.x for servers, and 192.x–255.x for infrastructure.

Reserve the first and last subnet in every block for future use. Never allocate the entire parent block at once — leave 20–30% unallocated for growth. Use consistent subnet sizes within each tier: all server subnets use /26, all user subnets use /24, all management subnets use /28. Consistency makes the plan easier to document, easier to audit, and easier to encode in firewall rules. Point-to-point WAN links always use /30 (or /31 per RFC 3021 on devices that support it). Loopback interfaces on routers use /32 and should come from a dedicated 10.255.0.0/24 range — they are not connected subnets and should be kept separate from transit and LAN space.

Document every allocation in a dedicated IPAM (IP Address Management) system — even a spreadsheet — before assigning any addresses. Undocumented networks accumulate quickly, overlap occurs, and the resulting troubleshooting is disproportionately expensive. For IPv6, follow the same hierarchical principle: a /48 per site, /64 per subnet, and document the plan before assignment. The topology simulator's auto-assign feature demonstrates the hierarchical /30 allocation principle — every link receives its own contiguous /30 block from the parent pool, with no overlaps and no wasted space between allocations.