Low Latency, High Performance Network Services for High Frequency Trading
Tue, 03 May 2011 11:16:00 GMT
An Interview with Alastair Kane
Sponsored by Level 3 Communications
In this interview for the High Frequency Trading Review, Mike O’Hara talks to Alastair Kane, Vice President of Segment Marketing for Financial Services at Level 3 Communications, the international communications company, about how their business is evolving to cater for high-speed, computer-driven trading.
High Frequency Trading Review: Alastair, can you give me a brief introduction to Level 3 and your own role in the company?
Alastair Kane: Level 3 was founded in 1998, is a Fortune 600 company with 5,500 employees, revenues of $3.9bn in 2010 and we’re one of the world’s largest providers of Internet connectivity. We are a fiber based supplier and the vast majority of our network is dug and operated by us, which gives us great control of our network at every level.
We are dedicated to moving large amounts of information and content around the world as quickly and efficiently as possible. Level 3 has an extensive network in both the US and Europe and we are constantly looking to expand into new markets. Level 3 is also one of the few carriers operating a Content Delivery Network, which is dedicated to streaming and delivering high quality video content. An example of this is the fact that we have delivered the live feed for the Superbowl for the last 22 years through a business unit that we acquired via acquisition a few years back.
Our expertise at delivering data at ultra-fast speeds means we have developed a close relationship with the finance community. We currently work with nine of the top ten banks and four out of the five top financial exchanges. These organisations value the fact that we have an understanding of the pressures that financial institutions operate under. In essence, speed and reliability are crucial to the success of trades, and we support this.
I’ve been with Level 3 for 11 years and I’m in charge of segment development in the EMEA region. My role is to make sure that we understand the finance community – and its ever changing requirements – to ensure we adapt our capabilities and put together really specific solutions to fit what the finance community requires today.
HFTR: The finance industry is constantly changing and each type of participant, particularly those involved in any kind of high frequency or high-performance computer-based trading, has its own set of requirements. What do you see as the key things these firms need from a network provider?
AK: First of all, speed. The ability to be first to the trade is clearly the competitive advantage that everyone is looking for whether you are a proprietary trader, an investment bank, or an exchange. Being able to provide low latency is crucial. I cannot underestimate how important this element is.
However, speed is nothing without a number of other crucial factors. The first of these is having a future plan. With the industry developing so fast, financial institutions need to be looking for a low latency provider who is forward looking enough to be continually improving their route times. There is no point in being the fastest route for a month You have to be looking at the long game. Continually improving routes is vital to providing a sustainable competitive advantage.
The second thing is reach. As trading becomes more distributed and people are looking to get into new marketplaces and seeking new access to liquidity, the ability to have low latency routes into multiple locations is really important. This allows traders a diverse range of portfolio options, spreading risk and increasing trading options.
We’ve been developing our low-latency capabilities across a number of markets, both within Europe and the US. We currently connect Chicago, New York, London, Stockholm, Frankfurt, Zurich, Milan & Madrid. We take a global view on low latency, as opposed to offering a narrow range of one or two routes. Certainly as the trading community looks to change the way that it trades (and this is an extremely dynamic space at the moment), this type of flexibility will be more and more important.
Finally, and this will resound with risk departments, is the need to be able to trust in your low latency provider. To offer service level agreements, full support and to have complete faith in the stability of your low latency provider is absolutely vital.
HFTR: What kind of steps are you actually taking to reduce latency on your routes?
AK: We have a team at Level 3 which is dedicated to reducing latency. This can be on a macro or micro level, the team know details of every route mile and when they identify a possible improvement they do everything they can to utilize this. This might only save fractions of fractions of seconds, but this matters to our customers in the finance sector. This is a process that never stops, at Level 3 we are always trying to make our low latency routes faster and faster.
Essentially, speed is increased by reducing the underlying route distance itself. In other words, how far does the data have to travel? A lot of the network that we have is our own fiber, so we understand both those routes and the latency on those routes. But we’re constantly looking at developing new routes. What we’ll do is actively test the physical latency, then we’ll look at the best possible theoretical latency and look for options to help us get closer to that number. We have the capability to dig. We’ve dug a lot of our own network and we’ll continue to do that where appropriate.
Where an improvement in speed can be brought about by interfacing with another provider, we will also do this. We own a variety of interconnect points where we can interface with complementary fiber in order to build faster overall routes, so this is an on-going development.
It’s also about the equipment. In terms of latency, the most important factor is generally the distance between the two points. But as you get better and better routes, the advantages that you gain are by their very nature reduced, because gradually you start to take out the redundancy. Therefore we’re investigating other ways to improve latency from an equipment perspective.
Overall, it is about understanding the entire picture, it’s about having control of your network and working with partners who you trust, so you can then translate that into services and ultimately speed for your end customer. We are very open about what our latency figures are and we back those figures up with SLAs between ourselves and our customers.
Certainly within the European market, we’ve seen a very beneficial reduction over the last 18-24 months in terms of latency and we’ll continue to see that.
HFTR: You mentioned equipment there. Can you give us some insight into what kind of equipment you’re working with to bring down latency?
AK: We are working on the very boundaries of what is physically possible and are occasionally even limited by science itself. If you look at basic principles of information transport, when you send a digital signal through any medium, it’s a series of ones and zeros, sent either as an electrical pulse through copper cable or an optical pulse sent as a laser down a glass fiber. About 15 years ago, it was discovered that one couldn’t turn the light on and off any faster than ten billion times (10 Gigabits) per second, so the only way to ’go faster’ was to send different colours of light down the same fiber. That is referred to as dense wavelength division multiplex (DWDM). It’s using a different wave in the colour spectrum. We’re currently deploying systems with 160 different colours of light, which means we can send 1.6 Terabits of data down a single fiber.
Level 3 then build on this underlying technology and customise the delivery mechanic for low latency trading. We can do this by using our own underlying algorithm which strips out non-essential processes, allowing data to travel at an increased pace. Now this may only make a fractional difference on a long route (between London and Frankfurt for example), but it makes a much more significant difference on a shorter route, say between two London data centres.
So that’s the sort of thing we’re doing, looking at the electronics components at the start and end of the signal path.
HFTR: I’d be very interested in any developments you’re seeing in other industries you operate in, like gaming for example, which might have relevance to high-speed, computer-driven trading.
AK: This is one of the core advantages that Level 3 has, because of our wide range of customers whose requirements are very similar, even though the businesses they operate in are very different.
If you look at gaming specifically, it’s a truly global industry, there’s a huge explosion of growth within Asia and across Europe, and the United States is still a massive market. But gamers are a highly distributed and highly demanding customer base, as well as highly educated in terms of the level of service they expect.
The gaming community is actually very similar in a lot of ways to the trading community. They want to be able to access the content that they need (in this case, the game) at the fastest possible rate and they want to be able to be as competitive as possible. The need to be able to beat your opponent to the punch is as prevalent as it is in trading!
At the same time, the requirements of the information that they’re sending is increasing all the time. In the past, you had standard definition games but you’re now moving to high definition & 3D games, which can be as much as 10x the capacity of the previous versions. So you have to have networks that operate in the places that they need to be, but which can also scale to take into consideration the growth in the amount of data they’re transferring.
Certainly within the financial community, increased use of video as an information source as well as huge amounts of market data mirrors that world, so for us it fits extremely well in terms of having customers across a range of business lines who have similar requirements.
HFTR: Level 3’s core competency seems to be running these high-speed networks that can deliver huge volumes of data. But industry participants – whether buy-side, sell-side or exchanges – are constantly evaluating different suppliers for other types of capabilities that sit on top of those networks. So what kind of services or products are you offering beyond the network itself?
AK: We’ve talked a lot about low latency, but we view this as a capability, as opposed to an end in itself. What we’re looking at is how we can help the trading community work however they want as effectively as possible.
We’re expanding our capabilities in a number of areas, such as providing the ability to trade locally where appropriate. In the UK for example, we have two proximity hosting sites within London, which are beneficially placed for the City, in Braham Street and Goswell Road. We’ve recently announced the opening of a third facility in Hayes, which is not only very conveniently placed between London and a financial exchange data centre in Slough, but it also has the backhaul for the transatlantic routes coming into it.
Again, this speaks to the point about creating flexibility of choice. Low latency is great for certain types of trading, but increasingly people want to be able to trade in a distributed environment. Companies must be able to understand this and to provide services that allow customers flexibility.
Putting the tools in place to allow people to choose and to be able to adapt over time is the key. Not to be locked into a single way of trading based upon your infrastructure, but to be able to make the choice depending on the changes in your business, not the changes in our business.
HFTR: Presumably some of your larger target customers already have substantial international networks in place. What are the issues they face when deciding what to build and what to buy when expanding those networks?
AK: “Build versus buy” is always a very interesting and dynamic question. In any industry it never stays the same at any given point because requirements change, networks grow and change, equipment improves and all of those dynamics are constantly evolving. And that’s just purely at the network level. Much more importantly of course is at the customer level, what are their requirements as far as their business plan is concerned?
It’s not as simple as looking at the cheapest alternative, there are many elements that come into play. What do you want your support infrastructure to be? What amount of change do you perceive in the services that you are offering over time? That may mean that you want a degree of flexibility within a particular solution and therefore might not want to tie yourself in one way or another. Or it may be a strategic requirement for you to build and maintain your own network, over and above any tactical decision that you might be making.
HFTR: Can you tell us anything about your commercial model? Are customers locked in for set periods, for example?
AK: Again, evolution, trust and integrity of our service are very important in our commercial model. If we are continually decreasing latency and offering a scalable and reliable solution then customers will not be looking to move.
This said, we also recognize that flexibility is beneficial to the industry. Being able to provide commercial flexibility, as well as continually developing the physical product is something that we want to focus on as an organisation.
One thing in particular that we try to focus on is understanding the different parts of the industry and different groups, so we can offer commercial terms that really will make a difference. A combination of the topology of our network, the amount of the network that we own and manage, the locations that we have, the physical infrastructure at a proximity hosting and colocation level and the transatlantic and US offering gives us the flexibility we need to deliver on this.
We understand that a proprietary trader has a different set of requirements to an investment bank or an exchange and we reflect this in our offering.
HFTR: Do you work with any specific partners or do you have a fairly open approach as to who you work with and how you work with them?
AK: We’re very open and select partners who let us deliver the best level of service. The majority of our network is Level 3 owned and operated, but we also understand the need to work with partners occasionally to deliver the best possible service. As mentioned, flexibility is key and we are not opposed to working with others where it delivers benefits to our customers.
HFTR: Finally, what do Level 3 have planned for the future?
AK: Faster routes in more and more places. We will continue to evolve our offering to the sector, grow the customer-base and work hard to keep our customers happy.
At the end of the day, that’s how we decide where we go. It’s on the back of what our customers need us to do and where we want to expand. There’s no doubt that we have two very good opportunities, at a geographic level and at an industry level, to expand, and we’re aggressively chasing both of those at the moment.
HFTR: Thank you Alastair.