OSPF scalability Test

Although Network Performance Analysis in enterprises and the access market has traditionally focused on data forwarding performance, the current and next-generation routers need to analyze the scalability of the OSPF protocol. Once the implementation of OSPF is analyzed in an isolated manner, the data forwarding performance of the router needs to be evaluated under the condition of stable and changed routing information.

Displays the devices to be tested in a medium-sized enterprise network. R1 is the backbone router connecting different departments. R1 forms an OSPF peer session with the other four routers, that is, the so-called adjacent adjacency ). Vror2 R2. Routers R3 and R4 are connected to R1 through an Ethernet switch. In addition, routers R1 and R2 connect to the Internet through R5. To accurately determine the performance of R1, we must be able to effectively simulate a medium number of adjacent routers and inject a large number of routes.

When testing R1, you must first be able to form multiple adjacent networks with R1. Each neighbor must be able to publish a route and run independently of other routers ., If you connect to a remote router through a second-layer switch, it is likely that multiple OSPF adjacent networks are formed at the same time and on the same port. Therefore, the test device must be able to simulate different independent routes on the same port at the same time.

The number of routes is also an important factor. Although most routing sessions publish a small number of routes, R5 may inject up to 0.1 million routes into the OSPF domain from the Internet. Because multiple connections from R1 and R2 are connected to the Internet through R5, R5 may choose to publish accessible information to R1 and R2, so that they can better make routing decisions on networks in the Internet. Although this additional information is useful, it often causes a reduction in overall performance.

In man, there are very different requirements. Many established network equipment manufacturers and a large number of emerging manufacturers are developing a new type of Router Based on IETFRFC-2547. The edge (PE) router of the provider is used on the edge of the service provider's Metro network to form an OSPF peering session with the enterprise router.

Because the operator's edge router is located in the network, it usually contains 128 Gigabit Ethernet ports. Each port is more divided into VLANs used by multiple users. A unique OSPF instance runs in each VLAN. A typical edge router PE supports thousands of OSPF sessions at the same time. Because each customer maintains its own route information, each OSPF session must maintain a unique database. This requirement also causes serious scalability obstacles to the use of traditional OSPF in enterprises. In a typical enterprise environment, vrouters maintain only one OSPF database regardless of the number of adjacent networks. When PE is used, an OSPF database is usually maintained for each OSPF neighbor.

When testing PE performance, you must be able to support up to 100 adjacent networks per port, allow each OSPF adjacent network to run in its own VLAN, and support a large number of ports in the overall system test. Because it is not cost-effective to dedicated independent CPU and memory to each OSPF adjacent, PE will share hardware and software resources between different adjacent areas. Therefore, the impact of changes in one VLAN on other VLANs must be evaluated. For example, if the status of a moderate number of OSPF changes in other VLANs, a significant performance degradation may occur in one VLAN.