| v6ops | Z. Li |
| Internet-Draft | China Mobile |
| Intended status: Informational | Q. Zhao |
| Expires: January 05, 2012 | X. Huang |
| Y. Ma | |
| Beijing University of Posts and Telecommunications | |
| X. Xu | |
| Huawei Technologies Co.,Ltd | |
| July 04, 2011 |
Analysis on Load Balancing Requirement in IPv4/IPv6 Transition
draft-li-v6ops-load-balancing-requirement-00
This document first analyzes the critical issues of bottlenecks in existing tunneling and translation technologies, which can be solved by proper load balancing mechanisms(e,g., scalability, availability, and single point of failure). Then, several key factors in designing a valid load balancing mechanism are described. Solutions to specific load balancing requirements can be drawn out by considering these factors. At last, current efforts about load balancing are introduced.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on January 05, 2012.
Copyright (c) 2011 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Since the depletion of IPv4 address, IPv6 is being introduced into existing networks and IPv4 will be gradually replaced by IPv6.
The Current Internet is almost fully built on IPv4. IPv6 is fundamentally incompatible with IPv4. The smooth and incremental transition from IPv4 to IPv6 is vital to IPv6's success. IPv6 transition technologies are generally classified into three catigories: dual stack, tunneling (e.g., 6RD[RFC5969], DS-Lite[I-D.ietf-softwire-dual-stack-lite]) and translation (e.g. , NAT64[RFC6146], IVI[RFC6219], BIH[I-D.ietf-behave-v4v6-bih]).
Dual stack technology can gradually steer traffic from IPv4 network to IPv6 network by using the address selection policy that generally "IPv6 prior to IPv4", as specified in [RFC3484].
Both tunneling and translation technologies introduce traffic concentration points, the tunnel end-point in tunneling technology and the translation gateway in translation technology. When the scope of IPv6 deployment continuously expands and the traffic of IPv6 business gradually grows, traffic concentration points usually bare very heavy work load. These concentration points will become bottlenecks and the sources of the single point of failure. To address such issues,load balancing mechanism should be considered.
This document firstly analyzes these problems caused by the centralized points imposed in the current tunneling and translation solutions. These problems are expected to be solved by proper load balancing mechanisms. Secondly, several key factors in designing such load balancing mechanisms are analyzed. Finally, current efforts about load balancing are analyzed and concluded.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].
The terminology used in this document is consistent with 6rd[RFC5969], NAT64[RFC6146], IVI[RFC6219], DS-Lite[I-D.ietf-softwire-dual-stack-lite]. Besides, the following terminology is defined.
Load-balancer: A system which distributes users’ workload across multiple devices providing identical services.
Target Device: The device to which a load-balancer distributes workload. That is the concentration points of the tunneling and translation technologies.
Selection policy: The Policy by which the load-balancer selects target device.
Transport protocol: The protocol the load balancer uses to delivery information of selected target device to user.
During the coexistence of IPv4 and IPv6, more and more IPv6 traffic will be carried on IPv4 network or IPv4 Internet in the form of 6in4 tunnel such as 6RD[RFC5969]. Similarly, more and more IPv4 traffic will be guided to IPv6 network or IPv6 Internet in the form of 4in6 tunnel such as DS-Lite [I-D.ietf-softwire-dual-stack-lite]. Both DS-Lite and 6RD concentrate all the tunnel traffic at the tunnel end-points in the service provider network. These tunnel end-points are traffic concentration points which are potential bottlenecks.
During the evolution from IPv4 to IPv6, IPv6 users will communicate with IPv4 network or IPv4 Internet in the form of 6->4 translation such as NAT64[RFC6146]. Similarly, some IPv4 users will visit IPv6 network or IPv6 Internet in the form of 4->6 translation such as IVI[RFC6219]. Both NAT64 and IVI convert all the translation traffic at the translation gateways in service provider network. These translation gateways are traffic concentration points or potential bottlenecks.
There are many problems about bottlenecks.
In order to avoid becoming bottleneck, traffic concentration point should be distributed to multiple parallel points. Each parallel point performs the same operations. There must be some load-balancing mechanisms to make parallel points collaborate with one another stably, smartly and efficiently.
In order to keep transition technology always available when traffic growing from small to large scale, some load balancing approaches should be applied at IPv4/IPv6 conjunction points. There are four categories requirements for load balancing scheme. Different solutions or mechanisms may be concluded when necessary to meet the requirements of each. The four categories requirements are: Target Device, Selection policy, Transport protocol and Load-balancer location.
As mentioned above, the target device can be tunnel end-point or translation gateway in service provider network.
In DS-Lite, such device is AFTR (Address Family Transition Router element) which is 4in6 tunnel concentration point and NAPT44 address port translator. In 6RD, such device is BR (Border Relay) which is only a 6in4 tunnel concentration point.
In NAT64, such device is stateful NAT64 device which is IPv6 to IPv4 protocol address translator. In IVI, such device is IVI translator which is the mapping and translation gateway.
[I-D.zhang-behave-nat64-load-balancing] has proposed some kinds of prefix64 selection policies for NAT64 load balancing. Those policies can also be applied to general load balancing selection policy in addition to Subscriber-Based Policy.
It requires a load-balancer to select target device according to whether target device and involved routers support anycast[RFC4291].
It requires a load-balancer to select target device according to the IPv6 address of user device.
It requires a load-balancer to choose target device according to e.g., the FQDN, the IPv4/IPv6 address, or other identifiers of destination servers.
It requires a load-balancer to select target device and destination servers circularly according to the arrival request.
It requires a load-balancer to choose target device according to the subscriber's information, which can be stored in AAA servers, for example.
It requires a load-balancer to select target device according to the real-time status of target device.
According to the selection policy, one or more target devices may be chosen by load-balancer. There are different protocols can be used to transport the results to user device.
DNS protocol carries response results to DNS queries. If result is IPv6 address with network specific IPv6 prefix, traffic can be lead to relevant translation gateway. If result is IPv6 address of tunnel end-point in service provider network, traffic can be lead to relevant tunnel concentration point.
DHCPv4 or DHCPv6 protocol sends network configuration parameters to user device. The parameters may contain network specific IPv6 prefix or tunnel end-point address.
ICMPv6 protocol can broadcast RA (Router Advertisement) message and options to user device. The options may contain network specific IPv6 prefix or tunnel end-point address.
RADIUS protocol[RFC2865]] can transport subscriber's information to network servers such as DHCP or DNS server. The subscriber's information may contain selected network specific IPv6 prefix or tunnel end-point address. The network servers then can distribute the prefix or address to user device.
The load-balancer can be implemented in different locations or different devices.
Terminal/Host can be modified to balance the load to target devices. The terminal/host should use the selection policy in 5.2 to choose a target device and make its traffic go through it.
Access gateway can be modified to balance the load on target devices. The access gateway should use the selection policy in 5.2 to choose a target device and make user traffic go through it.
Network servers include DHCP server, DNS server, RADIUS server and so on. They should use the selection policy in 5.2 to choose a target device. Then, they delivery the results to user device and user device will make its traffic go through the target device.
At present, some transition technologies have specified their own load balancing solutions, some transition technologies' load balancing solutions are in discussion and the others' are still waiting to be focused on.
Target Device is BR, the tunnel end-point.
Selection Policy is Anycast.
Transport Protocol can be DNS, DHCPv4, PPP, or manual configuration etc. DHCPv4 is recommended in [RFC5969].
Load-balancer Location is in network server. DHCPv4 server is recommended in [RFC5969].
BIH is described in [I-D.ietf-behave-v4v6-bih].Since BIH is NOT REOMMENDED to use together with a NAT64, there is no traffic concentration point for the BIH per se. So BIH does not need any load-balancing mechanism.
Target Device is the NAT64 translator.
[I-D.zhang-behave-nat64-load-balancing] is considering NAT64 load-balancing mechanism. But no solution is recommended at present.
Target Device is the IVI translator.
The load-balancing mechanism is not mentioned in the specification of IVI[RFC6219].
Target Device is AFTR, the tunnel end-point.
Although the load-balancing of AFTR can be done by using the FQDN option defined in [I-D.ietf-softwire-ds-lite-tunnel-option] and DNS, this point is not explicitly described in the specification of DS-lite [I-D.ietf-softwire-dual-stack-lite].
The potential security problem should be considered in the specific load balancing mechanism designed for the specific transition technology.
This document makes no request of IANA.
Note to RFC Editor: this section may be removed on publication as an RFC.
The following individuals have contributed to this document:
Dacheng Zhang
Huawei Technologies Co.,Ltd
KuiKe Building, No.9 Xinxi Rd.,Haidian District
Beijing 100085
P.R. China
Email: zhangdacheng@huawei.com