October 4, 2007 at 16:39Experts choose a new routing protocol for the Internet
Network World magazine has published a large article on the Routing Research Group within the Internet Research Task Force (IRTF), which is creating the next generation routing protocol.
The fact is that now the backbone routers use the Border Gateway Protocol (BGP). All routers must constantly update the host table and exchange these tables with each other. So, these tables have recently increased so much that there are problems with processing such a volume of information. Today, BGP tables contain information about 244,000 hosts, while a year ago there were 195,000 records, and six years ago there were 100,000. Although equipment manufacturers claim that their devices can handle 10 millionth tables, they are so fast the increase in the number of records still raises some concerns.
Another drawback of the Border Gateway Protocol (BGP) is the need to update IP addresses on the corporate network every time a company changes provider. If the development of the Routing Research Group succeeds, then this tedious task of network administrators will forever remain in the past.
The dual functionality of an IP address (host identification and its location on the network) will most likely be clearly distinguished into two parts.
The Routing Research Group has been working for many years, but recently it has dramatically stepped up its activities. The seriousness of the IRTF’s intentions is even indicated by the fact that six months ago they removed the former team leader of the Swedish professor Avri Doria from their post. Two people took his place at once: Tommy Lee, a specialist in routing systems, who worked for Cisco, Juniper Networks and Procket Networks (now he is back at Cisco) and Lixia Zhang, professor of computer science at UCLA University. Interested companies joined the discussion of the new protocol, including Cisco, Juniper, Alcatel-Lucent, Ericsson, Huawei Technologies, AT&T, BT Group and Arbor Networks.
Experts believe that the work of the IRTF is the most radical attempt to rethink the traffic routing system in the entire history of the Internet. Traffic routing and host addressing are two very close systems. Now, when the migration to the new IPv6 protocol occurs, it is time to upgrade the routing system as well.
There are many options available as the main alternatives to BGP, but two technologies have the most realistic chances: LISP and Six / One.
The Locator / ID Separation Protocol (LISP) was created by a team of Cisco engineers to separate IP address functionality into two parts: host identifiers and routing locators. The concept involves the installation of tunnel routers that will add LISP headers to information packets as they travel across the network.
The Six / One protocol was created at Ericsson and offers an alternative for IPv6 networks. With this approach, hosts receive fixed IP addresses that change only in the high bits, depending on which provider the host is currently connected to. These high bits are substituted automatically as soon as packets go through the new provider.
The IRTF hopes that all stakeholders can find consensus over the next two to three years. Whatever protocol is chosen as the standard, it will last at least five years before its commercial implementation.
The fact is that now the backbone routers use the Border Gateway Protocol (BGP). All routers must constantly update the host table and exchange these tables with each other. So, these tables have recently increased so much that there are problems with processing such a volume of information. Today, BGP tables contain information about 244,000 hosts, while a year ago there were 195,000 records, and six years ago there were 100,000. Although equipment manufacturers claim that their devices can handle 10 millionth tables, they are so fast the increase in the number of records still raises some concerns.
Another drawback of the Border Gateway Protocol (BGP) is the need to update IP addresses on the corporate network every time a company changes provider. If the development of the Routing Research Group succeeds, then this tedious task of network administrators will forever remain in the past.
The dual functionality of an IP address (host identification and its location on the network) will most likely be clearly distinguished into two parts.
The Routing Research Group has been working for many years, but recently it has dramatically stepped up its activities. The seriousness of the IRTF’s intentions is even indicated by the fact that six months ago they removed the former team leader of the Swedish professor Avri Doria from their post. Two people took his place at once: Tommy Lee, a specialist in routing systems, who worked for Cisco, Juniper Networks and Procket Networks (now he is back at Cisco) and Lixia Zhang, professor of computer science at UCLA University. Interested companies joined the discussion of the new protocol, including Cisco, Juniper, Alcatel-Lucent, Ericsson, Huawei Technologies, AT&T, BT Group and Arbor Networks.
Experts believe that the work of the IRTF is the most radical attempt to rethink the traffic routing system in the entire history of the Internet. Traffic routing and host addressing are two very close systems. Now, when the migration to the new IPv6 protocol occurs, it is time to upgrade the routing system as well.
There are many options available as the main alternatives to BGP, but two technologies have the most realistic chances: LISP and Six / One.
The Locator / ID Separation Protocol (LISP) was created by a team of Cisco engineers to separate IP address functionality into two parts: host identifiers and routing locators. The concept involves the installation of tunnel routers that will add LISP headers to information packets as they travel across the network.
The Six / One protocol was created at Ericsson and offers an alternative for IPv6 networks. With this approach, hosts receive fixed IP addresses that change only in the high bits, depending on which provider the host is currently connected to. These high bits are substituted automatically as soon as packets go through the new provider.
The IRTF hopes that all stakeholders can find consensus over the next two to three years. Whatever protocol is chosen as the standard, it will last at least five years before its commercial implementation.