RITE is funded by the European commission under the fp7-ICT programme.
RITE proposes to remove the root causes of unnecessary latency over the Internet. Whilst time-of-flight delay is inevitable, greater delays can result from interactions between transport protocols and buffers. It is these that RITE will tackle.
Historically the Internet community has worked to improve throughput and resource utilisation. We believe that the time is now ripe for research to focus instead on latency, because there are more and more applications and scenarios where low latency is critical:
- In Financial trading, a millisecond less latency is worth up to tens of millions of Euros per year to a major brokerage firm
- In multiplayer on-line games, increased end-to-end delays severely harm the perceived service quality
- In interactive multimedia services, like immersive video conferencing, quality and understanding is reduced if latency is high.
- In general interactive use of the Internet (e.g. Web), which typically involves sequences of numerous small objects, each one can get held up in buffers occupied by other traffic, resulting in cumulative delays of seconds rather than milliseconds.
The common feature of all the above is that latency matters. The focus on throughput at all costs is no longer appropriate – partly because the extra capacity being deployed means that increased utilisation is no longer the most critical research topic, and partly because delay-sensitive applications often need to send less than the network allows.
RITE takes a three-pronged approach to reducing latency — it includes both end-system and network-based approaches:
- End system – for example, we believe that protocol interactions cause extra delay and that protocols can be optimised to improve latency with no, or limited, impact on throughput
- Network – for example, we believe that large buffers throughout the network add to the end-to-end delay, and that they can be reduced or even eliminated
- The interaction between end-system elements and network elements. For instance where end systems can make latency-improving decisions by utilising network hints.
The mechanisms developed in RITE will be derived from the use-cases of financial applications, online gaming and interactive video. The mechanisms will be evaluated on specially designed testbeds focusing on each use-case scenario.
RITE solutions will work across the current Internet. There is evidence that excessive buffers are present in wired and wireless networks, as well as in hosts. Therefore all these aspects are in scope. We will focus on creating mechanisms for the future Internet that can also be deployed into today’s network without having to change whole segments of infrastructure. We will also identify where re-configuration can solve problems.
RITE’s activities includes:
- Problem analysis, including experimental research to locate the parts of the system where the latency increase is the greatest, and analysis of the nature of the mechanisms that induce delays
- Development of innovative end system mechanisms to reduce latency, including transport protocols and multipath optimisations
- Network mechanisms to reduce latency, including novel approaches for buffer reduction in routers and other network equipment
- Prototype implementations of these mechanisms, for example in Linux operating systems
- Evaluation — sufficient testing of RITE’s proposals to enable an assessment of them and for us to make recommendations for the industry
- Standardisation of the mechanisms – where this is necessary to allow for a broad adaptation of the project’s results. The consortium includes several IETF experts.
- Operational Guidelines – in some cases we believe it would be sufficient to issue advice on how best to tune parameters, through a suitable expert forum
By reducing network latency, RITE will contribute to giving European businesses a competitive edge in time- dependant applications like financial markets and online games. RITE research will also aid in reducing the waiting time for general Internet applications like short flows in Web 2.0 applications. By bringing the results to standardisation bodies like the IETF, and make end-host solutions available through operating system implementations, RITE will ensure the availability of the results to the European public.