The Evolution of IP
During the past year, parallels have often been drawn between the depression that’s recently befallen the telecom equipment industry, and the one that hit the semiconductor industry fifteen years ago. While many similarities have been highlighted, one that’s been generally overlooked is the manner in which each downturn has helped bring about a cadre of critics claiming that, due to inherent flaws, the basic platform that was previously driving the industry’s growth (then personal computers, now IP networks) will fail to live up to past claims about its ability to disrupt existing technologies. In the case of personal computers during the mid-1980s, the criticism revolved around both the PC’s ability to be user-friendly enough to appeal to the consumer mass-market, and robust enough to displace workstations and minicomputers in the enterprise arena. Today, IP networks are routinely assessed as being incapable of providing the reliability, security, scalability, and quality-of-service needed to fulfill the dreams of “IP anywhere” proponents, or at least not without the help of costly, older technologies such as ATM and SONET. However, just as the release of new technological platforms assuaged many of the concerns held during the early days of the PC industry, and allowed it to deliver massive growth over the following decade-and-a-half, the development of new, more advanced packet communications standards will likely play a similar role with regards to the proliferation of IP. The progress of one of these standards – IPv6 – is especially worth following.
Going back to the days when it happened to be a low-bandwidth, government-owned entity primarily used as a means for allowing academic researchers to share information, the Internet has depended on a version of the IP protocol known as IPv4. The server from which you happen to be accessing this article uses IPv4 to communicate, and – chances are – so do the routers and the PC you happen to be using to access it. Until recently, IPv4 was the IP communications platform of choice virtually all Internet-abled devices, and even now, most likely accounts for 98-99% of them. However, having been developed at a time when network requirements were far, far less demanding than they are now, and by government and academic researchers who couldn’t have possibly foreseen their lowly collaboration network growing into a high-bandwidth, global communications system encompassing hundreds of millions of users, the platform is beginning to show signs of strain.
Perhaps the most pressing problem faced by IPv4 going forward is that of address space (in essence, “phone numbers” for internet-abled devices): Since its address structure provides for a mere 32 bits of information, IPv4 can offer a maximum of approximately 4 billion unique IP addresses. Furthermore, misallocation of these addresses has resulted in hundreds of millions of them – if not billions – to go unused at the hands of various organizations and service providers. While dial-up Internet users don’t require permanent IP addresses (an addressed can be picked out of a pool, assigned by an ISP for the duration of a connection, and then reallocated once the user signs off), they are needed for broadband users, corporate LAN users, and most importantly, each PDA or mobile handset user with an always-on Internet connection. Already, in Japan and South Korea, the vast majority of handsets sold utilize such connections; and with the coming proliferation of GPRS and 1xRTT services, it’s quite likely that, within 2-3 years, the lion’s share of new handsets released will require permanent IP addresses. It’s expected that roughly 420-430 million handsets will be sold next year, a number that should rise meaningfully going forward. Clearly, it won’t be long until IPv4 runs out of breathing room. It’s for this reason that Nokia, among others, has been a staunch proponent of the IPv6 standard. Utilizing a 128-bit address structure, IPv6 possesses a theoretical limit of 340 trillion trillion trillion unique addresses (that’s not a typo), enough to guarantee that, unless we get into the habit of assigning IP addresses to individual atoms, the protocol shouldn’t run out of Internet phone number space anytime soon.
Beyond the address issue, there are other reasons why a migration to IPv6 would prove helpful in improving the general quality of IP services. They include the following:
1.) IPv4 requires that the provisioning of IP addresses to corporate LAN users that previously didn’t have access to an Internet connection must be done on a per-user basis, making this a highly tedious process. IPv6 will allow service providers and network administrators to simultaneously provision all users within a given organization/network.
2.) IP addresses are split up into two sections: One stipulating the network that a device happens to be using, and one identify the particular address of the device within the network. Since IPv4 was developed at a time when virtually all Internet-abled devices were permanently tied to a single network, difficulties – both with regards to manageability and performance - can be encountered when assigning addresses to mobile users as they roam from one network to another. IPv6 contains built-in messaging mechanisms to deal with this issue.
3.) IPv4’s ability to offer multicasting (simultaneously sending the same content to multiple users via a single transmission) was, in practice, limited due to scalability issues. IPv6 will be able to handle multicast transmissions to millions of users, something that may open the door to innovative IP video-on-demand services to cable network operators and mobile service providers. IPv6 also allows for a feature known as “anycast” transmissions. An anycast transmission allows for a transmission to be sent out with the aim of reaching one of several devices, with the device best capable of receiving it at a given point in time ending up the recipient.
4.) In order to alleviate processing issues that existed at the time it was developed, IPv4 was designed to occasionally break large packets of information into smaller ones while communicating over a network. However, with the increases in processing power that have been delivered over the past decades, this technique does more harm than good for a network’s performance, as the latency produced by the increased number of packets needed to be processed more than offsets the lower amount of numbers-crunching needed per packet. IPv6 prohibits the breaking down of packets in such a manner.
5.) Packets transmitted via IPv6 allow for a “logical sequence” to be assigned to them. That is, a device transmitting multimedia content to another IPv6-enabled node can make sure that it’s received in a particular order. This feature may play a significant role in improving quality-of-service for IP video and telephony calls. Also potentially benefiting quality-of-service is IPv6’s ability to allow packets to be prioritized based on the contents of their headers. Presently, such prioritization can only be done via more strenuous, higher-level processing.
6.) Network security can also be managed at the header level via IPv6, by means of a mechanism that can require a given device to enter login information to a server before being allowed transmit and receive packets from it. Once more, processing requirements are reduced significantly, this time with regards to the demands placed on dedicated firewall products.
Nonetheless, before IPv6 can be anointed as a savior of the telecom industry, or even a major benefit, certain issues still have to be resolved, the most pressing of which is backwards compatibility: While IPv6 traffic is capable of being “tunneled” over IPv4 networks, a number of interoperability issues still have to be resolved. Also, work needs to be done on guaranteeing that the migration of service providers from IPv4 to IPv6 is relatively smooth and painless in nature (never a sure thing when dealing with a pioneering new technology – just ask 3G equipment vendors), that the bugs are worked out of early IPv6-based network software releases, and that IPv6/IPv4 tunneling doesn’t result in significant problems related to latency and processing overhead. However, progress continues to be encouraging, with the protocol having been already integrated numerous products, ranging from mobile equipment from Nokia and Ericsson to servers from Compaq and Hewlett-Packard to routers from Cisco and Juniper; and while service provider adoption still remains limited at this time, a number of leading operators, including France Telecom, Telia, NTT DoCoMo, and Telecom Italia, have already expressed intentions to make network-wide deployments in the near future. Those invested in firms directly or indirectly tied with the telecom equipment industry – whether the mobile segment, the optical segment, or both – would be well-advised to keep a close eye on the pace at which both these firms and their peers adopt IPv6 going forward. |
