On co-location – data centre infrastructure and strategy


7 December 2016


Data centre infrastructure is the backbone of any financial services organisation. Whether it is a multinational bank or a small investment firm, many of the IT challenges are the same. We speak to Dan Young, head of IT operations and support department at UK investment house Killik & Co, about how it defines its data centre strategy.


Headquartered in the salubrious district of Mayfair in London, award-winning independent investment house Killik & Co has built a strong reputation for its private-client services and the close client relationships it builds. It opened its first office in 1989, on a site in Cadogan Street in Chelsea that used to be a pharmacist’s premises, and it initially served the needs of a small group of clients that included close friends and family. Since then, its desire to make the benefits of investing available to all has brought it great success.

The firm has been voted Wealth Manager of the Year many times not only by its clients, but also by readers of the Financial Times and Investors Chronicle. It has grown organically over the years, and has become a respected source of advice on wealth planning and investment services, including stockbroking, multimanager funds and managed investments. Since its inception, the need for a robust data infrastructure has grown and its reliance on technology is as great as any business in the industry.

“We have a large IT footprint for an organisation with 200 employees. Our IT requirements are as much as in my previous company, which employed three times as many people. We invest heavily in IT as a driver of the business and to improve client services,” explains Dan Young, head of IT operations and support at Killik & Co.

“The core of our IT infrastructure comprises the telephony system, the email system and the trading platform that we lease from Platform Securities. My role is to lead a team of six that ensures the IT infrastructure is available and running smoothly. Also, I evaluate and adopt new solutions where they are appropriate to the firm’s strategic aims. The aim is to support the activities of the internal stakeholders, which in my case are the investment managers, wealth managers and back-office team,” he adds.

It is relatively rare that Killik & Co makes a big investment in a new system. Its IT needs are relatively stable, but it does require flexibility and reliability as much as any other organisation. It has, therefore, a forward-looking approach to the requirements of the systems’ infrastructure in order to ensure optimal performance day in, day out.

“We review our IT platform every 18 months or so, to make sure that it is in line with our strategy. As the company is a partnership, we find that individual partners sometimes have their own specific ideas about what that strategy should be, so we need to review our situation regularly. We have recently created a new position – chief technology strategist – who sits between the partners and me, and whose role it is to define our technology needs because IT is such a strategic part of the business. We also have an internal systems development team that works on improving efficiency,” Young remarks.

“The key challenge we have to address with this team in place is the use of data. Our business relies on producing accurate and timely reports based on the data we hold. Some reports are internal, but others are external. We need to deliver these reports on a daily basis. Another challenge is information security. We have systems in place – firewalls, robust access controls and encryption of devices – but these can always be improved. Security has to be balanced with usability.”

Our strategy is pure co-location. All of our systems run out of two data centres. The production site is Global Switch 2 in east London and we have a back-up disaster recovery location in Manchester.

Partnerships underpin an optimal data centre strategy

Financial services is an industry rich with data, and it is from that data that the value of an organisation is derived. The data centre, therefore, plays a pivotal role in how well a business performs. Killik’s data centre strategy embodies the key concerns of any organisation in this industry – the need to balance accessibility with robust performance, reliability and security. Addressing those concerns is a question of finding the right partners.

“Our strategy is pure co-location. All of our systems run out of two data centres. The production site is Global Switch 2 in east London and we have a back-up disaster recovery location in Manchester. It is important to have that back-up site outside the M25. Apart from that, we have test and development capability in our Mayfair head office and in our Ipswich back office. Through a variety of providers, we have used co-location for many years. Up until 2008, we had an in-house data centre model, but we moved to co-location so that our team could focus on the core infrastructure needs of the firm,” Young says.

A co-location centre is a type of data centre in which equipment, space, and bandwidth are available for rental to retail customers. A co-location facility will provide space, power, cooling, and physical security for the server, storage and networking equipment owned by other firms. It will also take responsibility for connecting their hardware to a range of telecommunications and network service providers. In doing so, it will provide a cost-effective data centre solution and take out much of the complexity that goes with managing systems in house.

“Co-location has been very beneficial to us in terms of the overall experience,” Young adds. “Taking this route does limit you in terms of where your data centres can be, which depends on your managed network provider, and you have to rely on their back-up systems in terms of power, for example, and on their physical security measures. There are also lots of hoops to jump through in order to install your own hardware. It is always easier if your hardware is downstairs in your own office, but the gains from co-location far outweigh the benefits of in-house solutions.”

Since 2011, Killik has had its main data centre in Global Switch 2, which is the largest purpose-built data centre in Europe. It is a premium, carrier-neutral facility in London’s Docklands area, close to the heart of the city’s financial district. It has been built to an enhanced Tier-III specification and benefits from diverse feeds from the national grid that are considered the most resilient available to any infrastructure in the UK. It offers available power from 1kW to 8kW for each rack footprint, it has on-site diesel tanks to support-guarantee emergency power generators that can operate for 50 hours at full capacity and it has robust physical security systems in place.

“We rely on our partner Global Switch 2 to take care of cooling and power needs, for example, and maintenance and evaluation (M&E), and energy efficiency are things that come as part of our co-location strategy. But we are still conscious of space and power usage, which is why we moved from rack servers to blade servers. In doing so, we were able to reduce our footprint. We moved to Nimble as our storage vendor to increase capacity, and reduced the amount of power and space we required,” Young explains.

“When you are choosing whether to adopt a co-location strategy and when you are deciding which service provider to use, there are many things that you should consider. Firstly, think about the advantages and disadvantages of having your own data centre environment. Everything is easier when your hardware is within easy reach, but you must also be able to fully secure access to the server room. That was not practical at our offices. The second issue is how much control you want. Co-location means outsourcing some responsibilities such as redundancy of the power supply.”

In defining its co-location strategy, Killik chose services from its managed network provider. The decision was based in part on the desire to reduce complexity. It is easier to have one partner than to engage with another party to provide links into the co-location site.

“Global Switch 2 is about 40 minutes away from our Mayfair office, which is fairly convenient. When we visit our disaster recovery site, however, we need to travel much further, so we need to plan strategically in terms of what we want to achieve when we go to Manchester,” adds Young.

Failsafes, foresight and the future

The back-up data centre in Manchester is vital for the optimal operation of Killik’s systems and the delivery of its services. It is the key element in its disaster recovery (DR) strategy.

“We have a very robust DR plan and we test it regularly. In an ideal world, I would like to be able to ‘fail-over’ to our back-up site at a weekend and then switch back,” says Young.

Even data centres with the highest specifications can suffer outages. Uncharacteristically, Global Switch 2 suffered two brief outages in 2016. The first, which lasted 300 milliseconds, occurred in June and was caused by a lightning strike. The second, which lasted 222 milliseconds, was in September and was blamed on a high-voltage fault. It resulted in clients losing access to data and services for two days. In those instances, the facility in Manchester ensured that Killik was able to continue serving its clients.

One way is which Killik differs from some financial services organisations is that it does not face a huge challenge in terms of scalability at either its main data centre or its DR site. The firm has two full-height
racks in each data centre, which has provided plenty of capacity. Its choice of vendors and products has allowed it the flexibility it needs with the requirement to rip and replace server racks. In the future, a hybrid model using elements of the public cloud may become more important to Killik’s systems architecture, but, for now, it is up to larger organisations to explore that model more extensively.

 

“The business has stayed roughly the same size over the years, so we can easily switch to the back-up data centre. All of the services are in place. For now, it is purely co-location for us. We have some development servers that use the cloud. There are elements of Amazon Web Services (AWS) and Microsoft Azure, and we are certainly looking at where public cloud services are going. For now, we have no peaks and troughs of workload, so we are not an ideal candidate for bursting into the cloud,” Young remarks.

We have a very robust DR plan and we test it regularly. In an ideal world, I would like to be able to ‘fail-over’ to our back-up site at a weekend and then switch back.

 

“Public cloud services are not necessarily cheaper and you do lose an element of control. In ten or 15 years, we could see some organisations with no on-site data servers because everything is going into the cloud. That is the way the industry is going. But many companies will be using hybrid models for years to come, moving steadily further into the cloud over time,” he adds.

 

Though it may differ in scale to some of the financial services organisations that are considering their data centre strategy, Killik & Co certainly provides an insight into the benefits of not only co-location but also a data strategy based on a clear understanding of the unique needs of a business.

Dan Young has been with Killik & Co since September 2005, and is currently the head of the company’s IT operations and support department. His main focus is to ensure the availability and smooth running of the firm’s IT infrastructure.