What's your role at Sumerian?
As a Technical Lead, my role is to work closely with clients to help them solve a wide range of business and IT challenges through the application of IT Analytics. A key part of this involves developing new and insightful ways of conveying the results of our analysis in a clear, easy-to-understand manner, so that our clients can act quickly on our findings and get the results they're looking for.
What type of analytics have you been working on recently?
Recently I've been working with several global investment banks, helping them to optimise capacity and de-risk change by identifying business and infrastructure headroom hotspots within their highly complex pricing and post trade analytics environments. By taking a holistic, end-to-end approach to capacity and performance optimisation we are able to consider IT activity in business terms rather than just as ones and zeros. Through this strategy, we've been able to identify performance bottlenecks in a way that makes sense to the business and is actionable by operational and technical staff.
How is Sumerian's approach helping these clients?
Our clients operate in the most demanding, multi-tiered and often asynchronous environments where millisecond performance and on-demand scalability are paramount. Making sense of the huge amounts of data generated by these systems would be impossible without taking an analytical approach - mining and analysing the data to distil insight that matters. Only after spending significant effort in understanding our client's key business questions and true objectives do we begin the iterative process of devising the appropriate analytical approach.
Then, one of the most challenging aspects of my job is devising and applying creative reporting and visualisation methods that allow for targeted consumption of the insight we uncover by a variety of audiences, from the CIO down to technical specialists. At Sumerian, by combining proven statistical modelling techniques with novel, visual-metaphors we are able to provide clear answers to specific business questions.
Can you tell us more about some of these visualisations?
Capacity flowpipe
One of the major challenges faced by the CIO is how to standardise reporting and planning across different business applications. Each service will most likely utilise several different technology tiers all with their own load profiles and capacity peculiarities, not to mention different business processes and throughput metrics.To address this challenge we created the analogy of a water flowpipe, where business transactions flow from one end of the pipe to the other through the various technology tiers.
Throughout the pipe there may be various constrictions and expansions - analogous to capacity and performance limits, these illustrate the maximum flow rate. By qualifying the incoming business requests volume, we can quickly identify the maximum achievable flow rate at each technology tier and ergo highlight areas with lowest headroom. Behind the scenes we're generating enormous correlation matrices and hundreds of thousands of multi-variate regression calculations, but all our client sees is an easy to follow diagram highlighting high-priority areas for attention.
Distribution Intensity Map
One of the largest aberrations in IT Analytics at the moment is the dependence on aggregated values to provide a profile of system and business behaviour. The use of averages and peak values only serve to blur and distort the real picture, presenting an insipid danger embodied by what the public like to think of as 'lies, damn lies and statistics'. What's really needed is a method for aggregating data that doesn't interfere with the true story. This is especially true when dealing with multi-model, non-Gaussian data sets, such as is often the case with real-world data. To address this issue we developed the concept of a Distribution Intensity Map, which is essentially a 1 dimensional histogram utilising colour saturation as an indicator of activity. These provide the viewer with a much better breakdown of activity within a range, whilst being concise enough to allow a uniform comparison between multiple series.
Capacity Hotspot Indicator
Our clients' environments are hugely complex and diverse; as a result their infrastructure monitoring is extremely varied. The sheer size of these environments coupled with an often unwieldy range of KPIs can result in many tens of thousands of metrics, of varying but indeterminable degrees of usefulness in forecasting performance and capacity issues. Understanding how these individual metrics relate to changes on business load requires a strategy that looks not only at metrics at an individual basis, but also interrogates the way in which conjunctive movements are made. By utilising a Principal Component Analytic approach, we have been able to build a heuristically-tuned correlation and behavioural forecasting engine, which not only allows us to identify the load metrics most influenced by changes in business volume but activity monitors divergence from expected system behaviour. As a bi-product of this system we are also able to apply near real-time data to the model to determine current peak-load and future headroom for growth, and present all this in business terms.
Tell us about your background and interests?
Over the past 8 years I've worked with IT companies ranging from a tiny 'basement' firm to a 100,000+ employee global consultancy. Variety keeps me interested, and I'm lucky enough to have worked on engagements with over 20 global firms including investment banks, retail banks, professional services organisations and NPOs, as well as SMEs and start-ups.
My academic background lies in Artificial Intelligence, but my real leaning is towards business and entrepreneurship, luckily at Sumerian there is a great deal of scope for combining the two!
My life outside of work revolves around design and music with particular interests in print, visualisation and electronica.