Route aggregation, also called route summarization, is the practice of advertising one shorter prefix in place of several longer, contiguous prefixes that share the same leading bits. In routing, it lets a router describe a whole block of networks with a single entry, so neighbors carry smaller tables and see fewer updates.
For example, the four networks 10.1.0.0/24 to 10.1.3.0/24 can be announced as one summary, 10.1.0.0/22. Aggregation is a direct consequence of CIDR, classless addressing, and it is one of the main reasons the global Internet routing table remains manageable.
How route aggregation combines prefixes
Route aggregation works by finding the longest common bit pattern shared by a set of prefixes and advertising that pattern with a shorter mask. Every address covered by the original routes is also covered by the summary, so traffic still reaches the correct region of the network.
Forwarding stays correct because of longest prefix match: a router always prefers the most specific route it knows. Inside the aggregating domain the detailed routes remain, while outside routers need only the summary to send traffic toward the right next hop.
The table below shows how contiguous blocks collapse into progressively shorter summaries.
| Component prefixes | Number of routes | Summary route |
|---|---|---|
| 192.168.0.0/24 to 192.168.1.0/24 | 2 | 192.168.0.0/23 |
| 10.1.0.0/24 to 10.1.3.0/24 | 4 | 10.1.0.0/22 |
| 172.16.0.0/24 to 172.16.15.0/24 | 16 | 172.16.0.0/20 |
| 2001:db8:0:0::/64 to 2001:db8:0:ff::/64 | 256 | 2001:db8::/56 |
A clean summary requires that the component blocks are contiguous and aligned on a power-of-two boundary. The block 10.1.1.0/24 to 10.1.4.0/24 cannot be summarized as one /22 without covering an address range that is not actually present.
Benefits for routing tables and stability
The main benefit of route aggregation is a smaller routing table, which reduces memory use and lookup cost on routers. Hardware routers store forwarding entries in limited TCAM or similar fast memory, so fewer entries translate directly into lower cost and headroom for growth.
Aggregation also provides fault isolation. When a single subnet inside the summary flaps up and down, the summary itself usually stays stable, so the change is hidden from the rest of the network. Remote routers do not need to rerun their path calculations, and convergence becomes faster and calmer.
| Area | Without aggregation | With aggregation |
|---|---|---|
| Table size | One entry per subnet | One entry per block |
| Update volume | Every subnet change propagates | Changes stay local |
| SPF or path recalculation | Frequent across the domain | Limited to the affected area |
| Troubleshooting visibility | Full detail everywhere | Detail only near the source |
| Path precision | Optimal per subnet | Possibly suboptimal per subnet |
Aggregation in different routing protocols
Each routing protocol implements aggregation at specific points in its design, and the configuration point determines where detail stops. Link-state protocols summarize at area boundaries, while distance-vector and path-vector protocols can summarize on interfaces or at the domain edge.
| Protocol | Summarization point | Typical mechanism |
|---|---|---|
| OSPF | Area border routers and ASBRs | Inter-area range and external summary |
| IS-IS | Level 1 to level 2 boundary | Summary address on the boundary router |
| EIGRP | Any interface | Manual summary per interface |
| RIP version 2 | Classful boundary or interface | Automatic or manual summary |
| BGP | Autonomous system edge | Aggregate address with optional suppression |
| Static routing | Anywhere | One static route covering a block |
In BGP, the aggregate-address command can suppress the more specific routes or advertise them alongside the summary. The AS_SET option records the autonomous systems behind the aggregate, which preserves loop detection but makes the route change whenever any component path changes.
Designing an addressing plan for aggregation
Effective aggregation depends on a hierarchical addressing plan in which each site, region or area receives a contiguous, power-of-two block. Aggregation cannot be retrofitted easily onto scattered assignments, so the plan must be designed before addresses are handed out.
A practical rule is to reserve room for growth inside each block, for example assigning a /20 to a branch that needs only a /22 today. Wasted space is usually cheaper than later renumbering, especially with private IPv4 ranges and the vast IPv6 address space.
| Design situation | Recommended approach | Reason |
|---|---|---|
| New multi-site enterprise | One summary block per region | Regional summaries at the core |
| OSPF multi-area design | Contiguous ranges per area | Clean summaries at ABRs |
| Provider assigning customer space | Allocate from provider aggregates | Protects the global table |
| Multihomed customer | Announce specifics with care | Traffic engineering needs detail |
| Data center fabric | Summarize per pod or rack | Limits leaf and spine table size |
Risks, black holes and common mistakes
The classic risk of route aggregation is a routing black hole: the summary keeps attracting traffic for a subnet that has failed or does not exist. Routers usually install a discard route to Null0 for the summary, which drops such traffic instead of looping it, but the traffic is still lost.
- Overly broad summaries advertise addresses owned by another site and steal its traffic.
- Discontiguous networks under classful auto-summary, as in old RIP and EIGRP defaults, cause intermittent reachability.
- Suboptimal routing appears when the summary hides that one path is better for a specific subnet.
- Lost specifics in multihoming occur when an upstream filters long prefixes and only the summary survives.
- Forgetting the discard route on manual static summaries can create forwarding loops.
Aggregation is therefore a trade-off between scalability and precision. It works best at clear hierarchy boundaries, and it should be avoided where policy or traffic engineering depends on seeing individual prefixes, for instance with ECMP decisions or policy-based routing tied to specific subnets.
Put simply
Route aggregation is like a postal sorting center that only needs to know a city’s postal code prefix, not every single street. Mail for the whole city goes to one place, and local post offices handle the details.
Routers do the same with network addresses. Instead of remembering hundreds of small networks, a router far away remembers one larger block. The key benefit is shorter lists and fewer updates, while the cost is a little less detail about each individual network.
In practice
A university with a central campus and eight remote faculties gives each faculty a /20 block carved from 10.40.0.0/16, and every faculty splits its block into sixteen /24 subnets for labs, offices and Wi-Fi. The network runs OSPF with each faculty as a separate area.
The area border routers advertise one /20 summary per faculty into the backbone, so the core routers hold eight inter-area routes instead of 128. When a lab switch in one faculty reboots several times during maintenance, the core never recalculates its paths, because the summary stays up. The result is a backbone table that is about sixteen times smaller and noticeably faster convergence during campus outages.