Today's trading environment is increasingly characterised by trading strategies in which decisions are typically not taken by human beings, but by computers that are programmed to make decisions autonomously and in time periods that are measured in milliseconds. Traditional trading systems were fed data from a variety of secondary sources such as market data vendors rather than directly from the trading venue or exchange. Indeed, some sources were even 'tertiary', with data being fed from market data vendors who had received the data from other market data vendors or as contributions from market participants.
This 'cascading' of trading data via secondary sources has led to one-way latencies in the range of several hundreds to even several thousands of milliseconds - delays that are unacceptable in automated trading environments. According to the Tabb Group:
"Algorithmic trading and direct market access are the biggest disruptors in modern-day markets going from virtually 0% to 31% of institutional order flow over the past five years. To manage technology in this environment, firms need a massively reliable, scalable, and real-time platform with virtually zero latency as firms now have to cope with market data speeds of up to 70,000 ticks per second. To capture, analyze, and take advantage of data moving this fast latency becomes more critical as milliseconds really matter."
In addition to the requirement for low latency, the global trading environment is also being impacted by recently enacted regulations in major markets. New regulations such as Regulation NMS in the US and MiFID in Europe include obligations on best execution and trade-through of orders. Brokers and exchanges will now be responsible for demonstrating that they have executed their customers' trades at the best available price in the market. This will demand that the market data against which trade decisions are made is not only accurate, but timely as well. Executing a trade at a poor price simply because market data is delayed is not acceptable under these regulations.
BT Radianz meets the financial markets' demands for low latency access to market data and trade execution. We have designed our network for extremely low latency access to major trading venues and sources of market data.
Causes of latency
Within most networks, including an IP packet-switched network such as BT Radianz network there are four main causes of latency:
- Serialisation delay
- Propagation delay
- Network switching delay
- Queuing delay
Network delays are typically quoted as two-way latency figures (round-trip delay or RTD). However in most trading environments, market data is received via one communications path (e.g. direct from an exchange using multicast) whilst orders are sent back via a completely independent path (e.g. via a brokers' trading system). As such, one-way latency is the appropriate benchmark for measuring network delay for the financial markets.
Serialisation delay
This delay is caused by the time it takes to physically move data packets into a telecommunications circuit. Serialisation delay depends on both the size of the packet of data being transmitted and the physical circuit speed being used. For a packet size of 512 bytes, serialisation delays are:
Circuit speed One-way serialisation delay (ms)
| 64kbps | 64.00 |
| 1.5Mbps | 2.65 |
| 2Mbps | 2.00 |
| 10Mbps | 0.41 |
| 34Mbps | 0.12 |
| 45Mbps | 0.09 |
| 100Mbps | 0.04 |
From the above it can be seen that the one-way serialisation delays are significant at speeds of 2Mbps and below where they can contribute as much as 2 milliseconds (ms) or more. For many real-time trading applications that send trading instructions and receive acknowledgements, the above figures can be doubled.
BT Radianz uses very high-speed circuits in our core network to reduce serialisation delays between major financial centres. We also implement the fastest appropriate circuits between BT Radianz access points-of-presence (PoPs) and customers' sites, including exchanges and market data providers.
Serialisation delays in the BT Radianz Network are kept low because:
- The technical/business model of BT Radianz's extranet business is based on shared access infrastructure, resulting in many clients' connecting to network at T3/E3 circuit speeds and higher. These typically result in around 0.1ms of serialisation delay.
- BT Radianz connects major exchanges and ECNs into its network using high-speed connections that minimise serialisation delay. Many exchanges are connected to the BT Radianz network at 10Mbps or higher, resulting in serialisation delays typically below 0.5ms.
- BT Radianz uses very high speed circuits (typically over 45Mbps) to link major financial centres around the world resulting in extremely low serialisation delays throughout our core network.
Propagation delay
Propagation delay is the delay associated with the time it takes for an electronic signal to travel across the telecommunications infrastructure. This is mainly dependent on the distance traveled by the signal and is governed by the laws of physics. As a rule of thumb a figure of 0.6 ms per 100km can be used. So for two cities 500km apart, a one-way propagation delay of around 3ms can be expected.
In minimising propagation delay, it's the actual path that the electrons traverse that is important. Having a network topology that routes packets from London to Paris via Madrid is clearly not efficient and would result in a far greater delay than if the data packets go directly from London to Paris. In calculating distances, allowance must also be made for the fact that the 'as the crow flies' distance is not always the same as the fibre path distance. This is particularly true when there are large physical barriers (e.g. oceans) in the way.
If high-speed access circuits and high-speed data switches are used, the total network latency between any two sites that are not in the same city tends to be dominated by propagation delay. To minimise propagation delay, BT Radianz's network topology links major financial centres together using the shortest, most direct route.
The BT Radianz shared market infrastructure has been designed and managed for the financial community from day one using a 'flat' network topology. The major global financial centres of London, New York, Singapore, Tokyo and Geneva are all linked together directly and are at the centre of the BT Radianz core network. As an example, there is no need for traffic bound for Tokyo from London to go via any other city. This is true for any combination of city pairs from the above list.
Within a region the same approach is adopted. All major financial centres are linked to the major core regional trading centre. This again minimises the distance required to send data and so reduces propagation delay.
Switching delay
Another component of network latency is switching delay, which represents the time required to shift data packets through the various hardware components comprising a network such as hubs, routers, etc. BT Radianz utilises the latest, proven high-speed network technologies to minimise delay from these hardware components, such that switching delay within the network is typically less than a few microseconds (µs), a negligible delay when compared with the propagation delay.
It is a misconception that 'direct' leased lines are always faster than an IP packet-switched network. Telecommunications carrier networks have switching points and latencies that are comparable to packet networks and, all things being equal, they demonstrate similar latencies. Switching delays through high-speed packet-switched networks are of a similar order to those in leased-line networks.
Queuing delay
Another factor in network delay that is often overlooked is the need to correctly size bandwidth against the traffic profile of the circuit. In a trading environment, the actual bandwidth required is highly dependent on the application and can vary dramatically from minute to minute throughout the trading day. Ignoring this can result in queuing delay, a situation in which a data bottleneck develops because the circuit does not have sufficient bandwidth to support bursts in data.
For example, the traffic profile of a market data feed from an exchange exhibits relatively low data rates throughout most of the day, punctuated by spikes at market open, end-of-day, and immediately following the release of market-moving news stories. These data spikes can easily reach peak traffic levels that are three or five times the average if a significant market event occurs.
If sufficient bandwidth is unavailable for the peak traffic caused by these bursts of traffic, then data packets backup until the queue is cleared. Should the data queue become too great, packets are dropped and re-transmitted. In either case, the result is significant delays in the transmission of data. Invariably, these market data spikes - and the resultant delays - occur at times of market volatility, when timely data is most critical.
If the number of delayed or dropped packets is low, the slight increase in latency may not be readily apparent to market data displays or application monitoring tools; however it could considerably reduce the effectiveness of a programme trading application.
The BT Radianz shared market infrastructure is designed to operate at 40% of its maximum capacity so that, even in peak market traffic conditions, sufficient network capacity is available. In addition, our operational procedures include monitoring of customers' connectivity to ensure that sufficient capacity is available between the core network and customers' sites. Monitoring is constant and dynamic, so that capacity is added to the core network and customers are notified about increasing bandwidth requirements on a proactive basis before data traffic is affected.
Low latency access from BT Radianz
The BT Radianz shared market infrastructure has been specifically designed to provide the lowest latency connectivity possible. We also develop market-focused solutions that are tailored specifically to the unique requirements of finance industry customers.
Radianz Proximity Solution is an example of such a solution that can further reduce the market data and trading application latency for our customers. It enables institutional customers to have direct market access (DMA) to stock exchanges and trading venues around the world by combining BT Radianz connectivity to leading exchanges with our customers' automated trading servers. These 'black boxes' are housed in one of BT Radianz's fully-managed, world-class hosting facilities to provide customers with the fastest possible market data and trade executions.
