|Issue:||Asia-Pacific III 2013|
|Topic:||Myth Busting: Debunking Common Myths of IPv6 Adoption|
Tom Coffeen, Chief IPv6 Evangelist at Infoblox, has over seventeen years of production internet work design, deployment, administration, and management experience. In his role at Infoblox, Tom focuses on IPv6 architecture strategy and real-world deployment in the content delivery network and service provider space.
Tom previously served as VP of Network Architecture at Limelight Networking, managing a team of architects and engineers for all network architecture initiatives, routing policy, and vendor support for Limelight Networks’ routers and switches, supporting the global IP backbone and CDN (Content Delivery Network) server edge. He brings multi-year experience articulating IPv6 strategy and vision within the Internet IPv6 adoption community as well as to the public media.
Tom Coffeen is a graduated of Arizona State University.
Asian ISP should start migrating now, as Asia is running out of IPv4 addresses first, while the demand is accelerating. Some IPv6 myths can be dispelled: IP Address Management system is a necessity for inventory of 32 hexadecimal characters addresses; Services will not be disrupted, because ‘dual-stack’ will enable parallel operation for years; the changeover is not just a technology change – it will affect every business unit. IPv6 adoption plan is needed for the core, the edge and the corporate IT services. This must include upgrading OSS (Operation Support System) and training NOC (network operations center) staff. Large scale NAT/PAT is not a long term solution – it can break geo-location, prevent lawful intercept and limit TCP port availability (leading to broken applications for end users).
Let’s be blunt: the global depletion of IPv4 addresses is here and it has the potential to ruin service provider business everywhere, but especially in Asia. This is because the core business model of most service providers relies heavily on an inventory of freely available (and mostly free) public IP addresses. Therefore, it is potentially catastrophic to already squeezed profit margins, should such a critical low-cost resource vanish or become prohibitively expensive.
Now consider that the Regional Internet Registries (RIRs) responsible for allocating IP address resources are either running out of IPv4, or have run out already.
Asia’s RIR APNIC (the Asia Pacific Network Information Center) was the first to run out. They entered the final stages of IPv4 address exhaustion in February 2011. Now that they (along with Europe’s RIR, RIPE) are down to their last /8 (or approximately 16 million public IPv4 addresses), they are in emergency allocation mode: no requesting ISP can expect to receive more than a /22 of IPv4 address space (i.e., 1024 addresses). If Asian service providers haven’t already experienced the difficulty of obtaining new public IPv4 allocations, they should be prepared for extra scrutiny of their allocation requests now.
Such IPv4 depletion comes at a critical moment in overall Asian Internet adoption. A decade of unprecedented economic expansion has resulted in impressive Internet subscriber growth. China has 538 million Internet users for a total of 40 percent population penetration as well as 10 percent year-over-year growth in subscribers. India’s population is far less subscribed at 11 percent penetration with 137 million users but with a 26 percent year-over-year growth rate. The explosion in the number of Internet-connected mobile devices (many using more than one IP address) further increases service provider demand for IP address resources.
Without IP addresses to support existing and new subscribers, service provider business growth will slow or stop, operational expenses will increase (due to costly workarounds like large-scale NAT), revenue will shrink, and the service will suffer, resulting in lost customers and destroyed brands.
The message should be clear: service providers, particularly in Asia, must hasten to begin (or continue) planning for IPv6 adoption, and execute the plans soon.
But how should service providers approach IPv6 adoption? To ensure cost-efficiency, business continuity, and – most importantly – minimal customer impact, service providers should be wary of the following common myths as they plan for IPv6 adoption.
Myth One: “I’ll be able to manage my IPv6 address space just like I manage IPv4.”
The problem with this line of thinking has to do with the IPv6 address itself. The IPv6 address space is defined by a value of 2128, which equals 3.4E38 (or 340 trillion, trillion, trillion) possible addresses. The standard format of the IPv6 address uses 32 hexadecimal values and any standard host address typically incorporates the 48 bits of the MAC or hardware address.
While IPAM (IP Address Management) systems to replace spreadsheets have become more and more popular for IPv4 address management, such systems are absolutely essential for IPv6. As one might imagine, the operational challenge of an accurate inventory of host addresses that use 32 hexadecimal characters can be daunting.
The enormous size of the standard IPv6 allocation for service providers can cause confusion when technical staff remains mired in “IPv4 thinking”. Most service providers receive a /32, which is 296 (i.e., 7.9E28 or 79 billion, billion, billion) total addresses. With such a large space of a standard allocation, host address conservation is no longer a consideration, but service provider network architects will need to learn the new methods of IPv6 address planning, methods that greatly simplify design and improve operational efficiency.
Myth Two: “I’m putting off IPv6 because transitioning to IPv6 from IPv4 is guaranteed to be service-interrupting and costly.”
For nearly all service provider networks, it’s initially an adoption of IPv6 in the existing network. IPv4 will continue to operate within the same network infrastructure elements in parallel with the newly added IPv6 in what is known as a dual-stack configuration. Over time, IPv4 will be gradually decommissioned and de-provisioned, but such a transition from IPv6 to IPv4 is predicted to take many years, possibly a decade or more.
Such an approach reduces capital expenditures by leveraging the existing network to support IPv6. This presumes network infrastructure has been refreshed recently enough to take advantage of robust and improving IPv6/dual-stack features (very likely if the hardware in question is less than three to five years old). The dual-stack approach also minimizes the risk of service interruption leading to increased operational expenditures in the short-term.
Many service providers find that the legacy IPv4 network will need to remain in place to support custom back office and OSS applications that may never support IPv6 and will need to be gradually phased out and replaced.
Myth Three: “IPv6 is something the network engineering and network operations staff can figure out.”
Most modern companies of any size, including service providers, will have many business units or silos that interoperate to keep the business up and running. These silos within service providers can be organizationally isolated but all rely on both the selling of network services as well as the use of the corporate network and the services, applications, and processes they support to enable the business to succeed. IPv6 adoption has the potential to impact every business unit because as an addressing protocol it will touch everything that is network-connected.
Thus, it’s critical that any IPv6 adoption plan identifies the cross-functional nature of IPv6 adoption and defines stakeholders in every business unit to both understand and manage any actual or potential impact. For some silos, this will be no more involved than being a point of contact, as with any other cross-functional IT initiative. For other silos, there will need to be personnel handling service, application, or process remediation while working in concert with the primary IPv6 adoption team (usually a subset of network engineer, network operations or IT staff).
So, what is the right approach?
The exhaustion of IPv4 and resulting need for IPv6 adoption has the potential to be incredibly disruptive to service provider businesses in Asia. However, with the right approach, organizations can create an IPv6 adoption plan that works for their business. To find the right adoption plan for your service provider organization, the following points may help guide you in the right direction:
– Consider a phased approach to IPv6 adoption: IPv6 adoption is complex and has the potential to impact every service, application, and process that relies on the network. A phased approach can better contain this complexity and improve chances for a successful (i.e., low-impact, low-cost) IPv6 adoption effort. Common phases of IPv6 adoption for service providers can include planning and training, core network adoption, edge network adoption, and corporate IT network adoption. Many service providers will find that, after the training and planning phase, turning on IPv6 in the core of the network is quite manageable, given the typically good support for IPv6 on most up-to-date router and switch platforms. Given the increase in dual-stack and IPv6-only clients connected to the Internet (especially in Asia), it’s a good idea to make the service provider website available over IPv6 sooner rather than later.
– Make sure your NOC (network operations center) and customer support staff are trained in IPv6: Running a dual-stack network will bring a unique set of operational challenges and service provider network operations centers need to be up-to-speed on those challenges as well as how to successfully manage them. Customers and peers need to have confidence that the service provider is ready to properly support IPv6. All of the careful planning and deployment of IPv6 behind the scenes can be a wasted effort if NOC personnel sound clueless when customers or peers report IPv6 issues.
– Don’t assume Carrier-Grade or Large-Scale Network Address Translation (CGN/LSN) will solve all of the service-provider challenges around IPv4 exhaustion: The existing complement of service provider-ready NAT technologies, including MAP-E, DS-Lite and NAT-PT, DNS64/NAT64, may prove useful in extending the life of IPv4 for some service providers, but be cautious. The economics behind the viability of such solutions are not necessarily well understood for the long-term. Be aware also that CGN/LSN solutions can break geo-location, prevent lawful intercept, and limit TCP port availability (leading to broken applications for end users).
– Start now: Given the alarming data around IPv4 exhaustion mentioned above, it should go without saying that if you haven’t started planning and executing IPv6 adoption, you’re already behind. The good news is that there is more practical information, peer operational experience, and vendor support for IPv6 adoption than ever before. Don’t be afraid to make mistakes as you gain operational experience of your own.
With execution of the right IPv6 adoption plan, service providers will help maintain the continued growth necessary to protect margins. Additionally, IPv6 adoption will increase network readiness for applications leveraging IPv6, while preserving business agility and helping to future-proof your organization as the world becomes ever more connected.