The internet has of course been wildly successful over the last thirty years as more and more functionality has moved online. A large part of this success has been due to two key protocols that have allowed the internet to scale relatively gracefully: HTTP which stands for Hypertext Transmission Protocol, and DNS which stands for Domain Name System. HTTP is the protocol used to send data between a web browser running on a laptop or mobile phone and the web page or application that it is communicating with, which is running on a server in the network. No matter where the web page is located or who develops the web browser, it is guaranteed that they will be able to interoperate because they all use the standardized HTTP protocol to communicate. DNS is equally fundamental as it is the protocol which allows end user devices to translate a given human readable URL such as “www.google.com” to a machine usable IP address that the network can make sense of.
When were these protocols introduced
HTTP and DNS are defined in the Internet Engineering Task Force (IETF) standards group. The original HTTP standard was published in 1999, and roughly parallels the time when the number of web pages was starting to grow exponentially. The original DNS standard was published even earlier in 1987, as it was required in the pre-web browser days for other applications such as for translating email addresses to IP addresses.
Since these early publication releases, there has typically been a new revision of the HTTP and DNS standards every three or four years to add incremental features such as improved security or robustness. However, in the last few years, the update mode in HTTP and DNS has rapidly changed from rolling out incremental features to more frequent and more major evolutionary steps. So what is in the future for these two critical system technologies?
Reshaping to support the 5G vision
Many technical reasons are driving the changes in HTTP and DNS. However, at the highest level, the main driver is certainly the rapid evolution of the internet architecture to the virtualized model. In the last few years, we have seen the migration of many of the internet’s applications from standalone physical web servers to virtualized platforms located in immense centralized data centers. Looking ahead we can see that 5G networks, which are expected to be deployed in the 2020 timeframe, will take this to the next level creating new requirements for the evolution of HTTP and DNS.
The NFV and MEC trends are driving major changes in the HTTP and DNS protocols. HTTP will need to become more streamlined and lightweight to meet the high throughput and strict delay requirements of 5G. Some key protocol development has already started in the IETF in an effort called QUIC, which stands for “Quick UDP Internet Connections” for HTTP. QUIC runs HTTP directly over UDP with a thin shim layer for encryption and flow control. This makes it run much faster than the traditional HTTP over TLS security over TCP, which has multiple layers of handshaking and encapsulation. The main use case for QUIC is smartphones accessing content over wireless networks where the traditional HTTP/TLS/TCP has well known performance problems which frequently requires mobile operators to put in a middle box called a TCP optimizer to improve mobile web browsing experience.
Similarly, for DNS, the expected roll out of billions of IoT devices with 5G connectivity will create whole new requirements for discovery and addressing of these devices. Again, some key protocol development has started in the IETF in an effort called DNS-SD which stand for “DNS Service Discovery.” DNS-SD allows quick discovery of local devices and services by having all the devices multicast to each other in a peer-to-peer fashion. In a residential IoT setting, this will allow the light switches, for example, to discover all the light fixtures without requiring any human configuration or management steps. DNS-SD thus will scale much more efficiently in the future than the traditional DNS approach which requires centralized query servers and much manual configuration.
When will we see these new protocol enhancements in our devices?
A lot of the attention on 5G is quite rightly on 3GPP where the key radio interface specifications are being developed. 5G will, however, be much more than a new radio interface. IETF is working on the protocols and application support that will complete the 5G technology stack. Protocols like QUIC and DNS-SD are just two key examples. It is interesting to note that, while we are all waiting in quiet anticipation for the 5G radios to be defined, 5G networking protocols like QUIC and DNS-SD will be available commercially before the 2020 timeframe of initial 5G systems. In fact, some of these protocols will be out commercially well before the 2020 timeframe. For example, QUIC is already deployed experimentally by Google in millions of their Chrome browsers on smartphones, laptops and Google servers. Most people do not realize that when you choose to download Chrome you are actually signing up to be a “tester” for Google and some of these new protocols. The performance results of this internet scale experiment is regularly feedback into IETF to allow for rapid protocol development. This allows the IETF to set a target final protocol release date of 2018 for QUIC, certainly a few years before we will see any final product release grade specification for a new 5G radio.