IETF wants to solve scalability problems in Internet routing systems

IETF is setting up a new working group to solve the scalability problem in the Internet routing system. This problem is caused by the Division of network traffic between various operators. The Working Group will be built on a proposal from the Cisco engineer team to build a new tunneling mechanism for Internet edge and core routers.

This new approach, known as LISP (locator/separator protocol), aims to reduce the number of route table entries stored in the core router controlled by Internet operators. LISP logically separates IP address blocks. Companies will use edge routers to promote their two major features to the global Internet: one is to determine the systems that use these IP addresses, the other is used to locate the locations where these systems are connected to the Internet. This operation allows LISP to collect location information, so generally it does not need to be stored in the core router.

LISP runs in dynamic packaging. Each data packet that enters the core route gets a new IP packet carrying the network information of the target service provider, rather than the IP address of the end user. This IP packet is removed from the packet when it reaches the target service provider.

LISP can be shared with the Border Gateway Protocol (BGP), which is the main communication channel between the edge and the core router.

Cisco researcher Dino Farinacci believes that the problem we face is that IP addresses are allocated to hosts and they are not allocated in topology mode. This means that the core router of the ISP network must carry all the specified routes.

LISP advocates believe that this technology also makes it easier for some companies to implement carrier conversion without using new IP addresses. Because the recognition function remains unchanged, even if the location information is changed. While LISP provides enterprises with additional transmission engineering capabilities.

Farinacci added: "More and more companies want to build multiple household addressing technologies for their own web pages, but they also hope to be at a low cost. Now they have to spend a lot to make this happen. They need to use BGP and release the path to the core router. With LISP, we place the routing policy on the edge router, and the customer can control the bandwidth on the edge router ."

Cisco engineer Darrel Lewis believes that the key aspect of LISP is how the web page can negotiate with the multi-host connection policy in an independent and open manner.

LISP developers think that the protocol will be deployed as software for edge route upgrade, but there is no requirement for hardware upgrade during the operation. They think this can be gradually expanded and used together with the current Internet protocol IPv4 or the long-awaited IPv6.

The LISP developers have been engaged in LISP for two years. The main LISP documentation is available in its 12th versions. Cisco engineers stressed that they would not launch any LISP-related products at this stage.

The LISP developers expect the LISP Working Group to be recognized by the IETF this summer and will continue to work on prototype and Protocol improvement.

LISP also has many opponents. Some conference participants in LISP think that this approach will increase the complexity of the core router because it adds surveying and tunneling services. The IETF leadership team has made it clear that LISP is a test. The working group claims that LISP has a potential negative effect on Internet transmission. At this stage, it is not recommended to deploy LISP in areas beyond the scope of the test.

Route table Growth

LISP is a by-product of IETF's sister company, the Internet research team (IRTF), and is intended to rethink Internet routing infrastructure. Both IRTF's routing research workgroup and LISP share a common goal: to reduce the size of BGP Route tables.

The BGP Route table is the master list of the network target. It is saved in the master router to determine the optimal path from one network to another. Experts are worried that increasing BGP Route tables will load the entire process and the memory requirements of Internet core routers. The increase in BGP Route tables is important because it requires the carrier to use a larger capacity router, which increases the carrier's cost. A key factor driving the growth of Route tables is that the company adds multiple addresses to improve the reliability of its network. The BGP Route table has 0.288 million entries and is growing at a rate of 14% per year. About 26 thousand enterprises have processed multiple addresses for their networks. There is no statistics showing the growth of Route tables. However, once such a problem occurs, it will be a disaster that cannot be solved by our current practice.

Now, all core routers are aware that all route tables or endpoints may have problems. With LISP, the core router only needs to understand the routing information required by the current data stream. LISP theory sounds good, but in practice it creates a new network that is totally different from our current network. Tunneling technology increases funding and slows down network traffic. LISP changes the time and cost. LISP relies on tunneling technology, which is not perfect. Sometimes, data is even lost because the transmitted data is too large.

1. Huawei released China's first RFC standard in the IP routing Field
2. The IETF edge router solution facilitates the transition to IPv6