As competitive and regulatory pressures mount, financial institutions are facing an unprecedented squeeze on costs -particularly in operations and technology. At the same time trading applications have become more complex, making it increasingly difficult for operations teams to fully understand their technology.
So on Thursday 15 March we held a webinar to discuss how application performance monitoring can make your trading infrastructure more profitable -while keeping costs down.
Even after years in the business, I was slightly wrong footed by what emerged. According to a poll of attendees during the webinar, it seems that the biggest obstacle to a profitably functioning trading infrastructure is not technical. It is human.
Yes, people are the weakest link when it comes to the pursuit of ‘perfect flow’. Failure to relate monitoring to business objectives; failure to communicate monitoring information to business users; failure to accurately understand the level of sophistication of the monitoring required and the failure to fully comprehend how monitoring can benefit the business, all emerged as important obstacles. (See chart below)
| Webinar Poll Questions |
Yes |
No |
| Do you expose any monitoring information to your business users? |
41% |
59% |
| Does the business use your operational reports to manage the business on a daily basis? |
20% |
80% |
| Have you been disappointed by the apparent level of sophistication of the monitoring required by the business? |
60% |
40% |
| Has the apparent level of simplicity led you to underestimate the scope of the monitoring required for the business? |
77% |
23% |
|
|
|
In addition, we wanted to address three key questions:
- How do you develop real-time flow monitoring across the trading infrastructure?
- What metrics do you need to monitor in order to improve profitability?
- How can business aligned monitoring reduce operational costs?
Mark Reece is a consultant specialising in business activity monitoring, particularly latency and performance. He was formerly an eTrading architect at HSBC and the London Stock Exchange.
He argued that IT often provides the business with complex monitoring data, but in reality the business doesn’t get it. All the business cares about is the impact on business.
He cited the example of a bank where order and quote management systems were being compared and the latency figures varied widely. At the same time a trading desk was being arbitraged by aggressive price takers. High latency was blamed and an expensive IT fix was planned. But by looking at the underlying timestamps of the price- taking trades and the bank-generated orders and quotes, he discovered that latency wasn’t the problem. In fact the price calculation system was running slow.
He concluded firstly that you have to engage the business to discover what they use to determine if a trading strategy or process is working.
He reminded us that even alarming findings from APM only matter if they affect business performance. Only the business managers will know, so you have to ask them.
Finally he pointed out that business managers aren’t really interested in monitoring –they just want an easy way to tell when things are going go wrong.
This was confirmed by John Crackett, CEO of Parallel Thinking, a specialist designer of banking and payment software and former Director of Technology at PLUS Markets.
John explained that at PLUS Markets the apparently simple question; “Are we ready to open?” meant much more than knowing that the technology was available. He had to know that the static data updates had been processed, that the market data vendors were connected and able to receive prices in order to run a transparent market.
He needed a platform that could examine technology, applications, transactions and external interaction and bring all this together. What he had was a collection of separate monitoring tools and applications which pulled in so much data that he couldn’t make sense of it.
He described how it took him months to understand the business sufficiently to reduce the number of key metrics from 60 to 3 and develop a simple red/amber/green traffic light system. (He said Geneos would have done it much faster.)
But he warned that simplification must be done carefully: trading parameters are dynamic and change over time so either you have to repeatedly go back to the business or better still, set up self-learning monitoring rules that change over time.
Kevin Covington, CEO ITRS Group
