Rules for assigning IPv6 addresses 2

previous 2/2 next

Procedures after obtaining addresses

User companies obtain IPv6 addresses from their ISPs, and information regarding those addresses will be registered with whois and other related databases in a manner similar to IPv4 addresses. In the following section, we will go through procedures you must perform after obtaining IPv6 addresses.

(1) Registration procedures and operation of whois

Whois database is operated by RIRs such as APNIC/ARIN/RIPE NCC. When addresses are first assigned to ISP, those addresses will be automatically registered with whois based on the information contained in the address request form. However, all the subsequent updates to this information must be performed by ISP, who is an address holder of these addresses. ISP directly changes the entries in the database operated by APNIC. When ISP assigns addresses to a company, it must register the information regarding those addresses with whois database. Since information regarding your company will be registered and will be referenced globally, we recommend you to obtain explanations regarding the whois registration information from your ISP.

A proposal for linking the whois database operated by APNIC with those operated by JPNIC in the future has been made. However, in the case of IPv6, various procedures and detailed rules regarding the actual operation of the database including, the level of user who needs to register, guidelines regarding the disclosure of registered information, and how to manage the registered information has not been well established yet. Issues regarding linking the registration procedures with higher name server, is also under discussion.

The manual for APNIC’s whois has not been translated to Japanese yet, but there is a plan to incorporate whois into IRR (Internet Routing Registry), so we recommend you to check the latest information frequently.

(2) Reverse DNS

APNIC also operates reverse DNS for IPv6 address. Therefore, after the allocation of addresses, registration with higher name server must be done directly through APNIC.

Actual registration procedures are on APNIC’s web site (in English), and there is also a Japanese translation on JPNIC’s web site (http://www.nic.ad.jp/ja/translation/apnic/apnic-reverse-dns.html).

Check with you ISP regarding the procedures for registering with name server, and follow their instructions.

(3) Addressing plan after the address assignment

In IPv6, unlike IPv4, most services automatically assign /48. Therefore, the amount of address space consumed will not become a big issue when designing corporate network. Here are few points to keep in mind when managing the allocated addresses inside the company.

• Be efficient when distributing addresses
  Try to be efficient not necessary in terms of conserving the address space, but in terms of dividing addresses in a way that allows more efficient use of the network. One of the things to consider is the aggregation of the internal path (where do they aggregate, and what happens to the total number of path.).
  
  
• Divide address block into parts
  You can divide the obtained address block by physical factors (for example, by regions such as North Japan, Kanto, and West Japan), or by logical factors (for example, by departments, business units, or other organizational units). Priority should be placed on usability and manageability.
  
  
• Deciding on address prefix
  Assign a bundle (for example /50 per bundle) composed of several units (for example /64 per unit) to each parts, and decide operation rules in a way that smaller divisions gets assigned a /64 from that bundle.

The above descriptions assume that there is only one upstream connection, and a single /48 is divided inside a company. When each department in a company independently applies for IPv6 network service, multiple /48 will be obtained. However, since all addresses are global, there is no problem with mutual connection as long as you take care of the security issues properly.


Latest information on policies

(1) Site-local address

One of recent important movements is the debate at IETF regarding the use of site-local address.

Some of the people who claim there is no need for site-local address has proposed a “Global link local unique address” as a replacement for site-local address. People supporting site-local address also have several alternative plans, and in the middle of April this discussion has moved to a vote for the appropriate plan. There is also an argument stating that these types of issues should be discussed at the registry level rather than at IETF, and it seems likely that registries will also be involved in the discussion from now on.

(2) Multihoming

In IETF, working group called v6Ops is discussing the issue of multihoming in IPv6. Originally, IPv6 was built on the policy of “Provider Aggregation” (aggregating the path at the service provider level), and was not considering a situation where an end user site has multiple connections with the same address as in the case with IPv4. Therefore, the direction of this discussion will be critically important for many aspects of IPv6, including traffic control.

In this article, we have looked at procedures for obtaining IPv6 address. However, it seems that registry does not yet have an answer to some complicated issues such as what to do when a user wishes to do multihoming in order to increase the redundancy.

Also, issues such as, the level of user eligible for direct allocation from RIRs, and disclosure and management of registered information, are expected to become a topic of debate as the situations change in the future.

As we have seen in this article, IPv6 address assignment procedure is relatively well organized at this point, but the details are still under discussion. Therefore, we recommend you to check the latest information in a timely manner.

previous 2/2 next