A recent report in Bloomberg quoted Goldman Sachs analysts predicting that volatility in US markets will spike after election day. Their conclusions, based on options activity, back up a feeling that many share on Wall St: volatility is high and only going to increase as we move deeper into the fall.
At Turbonomic, we’re seeing indicators of similar volatility, but from the IT perspective. What’s interesting is that these indicators reinforce and amplify other dynamics to increase volatility around Fintech/Fiserv IT greater than volatility in the markets themselves.
Volatility in the Near-Term
The Bloomberg report quoted the Goldman note to clients: “Option markets seem to have abandoned the view that volatility would rise strongly in the lead-up to the election, which had been priced in throughout much of 2020,” Goldman wrote in a note to clients. “Instead, currently markets are expecting volatility to jump at Election Day, and then remain high thereafter.”
A look at a chart on VIX futures published by Bloomberg illustrates this market dynamic:
In order to prepare for this volatility one of Turbonomic’s largest banking customers for the first time is considering enabling automatic “size-up” actions for production workloads. The reason they’re moving to automation is they understand that they might experience unpredictable demand on some of their applications, and with over one-hundred thousand workloads to manage, being able to respond to performance issues caused by sudden spikes in demand isn’t something they can expect their IT to do manually anymore using a spreadsheet.
Another leading money center bank is approaching this challenge from a different angle. Rather than automating resize actions (an initiative it had already begun), it is accelerating its move to “super clusters” – increasing the size of their virtualized clusters beyond existing vendor limitations – to better distribute the demand driven by greater volatility across larger, more resilient, resource liquidity pools.
A third bank is putting an emergency plan in place to automatically shut off all development VMs in case production demand peaks beyond the capacity of infrastructure supply – we’ve seen that most large organizations are managing clouds with both development and production workloads sharing infrastructure. Obviously, development VMs can “take five” while waiting for production demand to subside.
Compliance and the New Normal in the Longer-Term
Compounding the various challenges around near-term market volatility faced by IT in the Fintech space, financial IT professionals also have to deal with volatility from navigating Covid in the medium term (and how that will transform business in the longer term), as well as new compliance requirements from regulators.
The Covid-19 pandemic and the resulting shift to tele- and remote work have changed a century’s old paradigm centered around the office, leading to increased demand for IT services to enable teleworking – and increasing stress upon financial IT organizations.
We won’t discuss the over-written example, Zoom (or whatever online collaboration technology your organization has chosen) here, but let’s consider a less-discussed topic: Virtual Desktop Infrastructure, or VDI. Considering the multitude of security and compliance requirements imposed upon financial institutions, the expansion of VDI deployments was a logical result of a world adapting to a pandemic. While VDI is incredibly flexible and feature-rich in replicating the highly complex desktop environments financial institutions must provide to maintain compliance and security, that infrastructure comes at a significant cost – namely capacity.
If not properly managed, expanded use of VDI will cause performance congestion. Significantly, those issues may amplify demand spikes caused by market volatility; VDI, after all, shares many of the same infrastructure resources as customers do. So, if, for example, a major market-moving event occurs, you might not only be facing a demand spike on customer-facing resources but from those dedicated to VDI as well – if an event catches your customers’ attention, it has likely caught that of your employees. Without a system that can self-heal to redirect resources automatically to overcome such spikes, financial institutions risk performance degradations at best, and potentially outages in a worst-case scenario.
In addition to increased demands of the new normal come demands from regulators – in particular, regulations from the Federal Reserve requiring financial institutions to prove they have a disaster recovery (DR) solution in place to maintain business continuity.
In practice, this means banks and other financial institutions must have a solution in place where they can lose an entire data center without missing a beat. Just as importantly, this applies not only to on-premises data centers: In some cases, for customers of AWS, Azure or GCP, if they have an application running on a public cloud, they need to prove they can also run that application on prem or on a different cloud. This isn’t as simple as having redundant capacity and data replication in place, because the cost of simply duplicating Fiserv infrastructure resources for every app would be astronomical. Furthermore, it impacts the ability to use a cloud specific service when designing these applications, and requires a system that can consume data from all private and hybrid clouds and apply decisions based on the entire estate.
App Prioritization is Key
The key to any solution for market volatility in the short term or increased requirements in the long term is application prioritization. Not all apps are created equal or have equal demands at the same time.
Prioritization can enable you to get resources to those customer-facing apps stressed by increased volatility. While developer- and test-facing applications can suffer brief periods of performance degradation, customer-facing Fiserv / Fintech apps cannot. How does your institution identify those apps facing demand spikes and reallocate resources in real-time today? Are you prepared for rapid swings and surprises?
Just as important as giving a customer-facing production app priority, a resilient system must give that app the resources it needs – but no more and no less than needed. In a rush to respond to performance degradation caused by an unexpected demand spike it isn’t surprising to see resources over-allocated, after all, no one will fault a fire fighter for flooding a fire; however, IT professionals know that in a fire the resources they have to firefight aren’t as unlimited as water.
An intelligent system that not only understands precisely what apps are suffering congestion due to volatility, but also exactly what those apps need, is necessary for overall system resiliency. That same intelligence must also understand when the demand fades without delay, so resources can be freed up to prepare for the next spike in demand.
Such an intelligent understanding of application resource management can help with regulatory compliance, such as Fed regulations regarding systems resiliency. A granular, real-time understanding of the equilibrium between application demand and infrastructure supply, and all the tradeoffs among resources like CPU, memory, storage, network and the like necessary to maintain that equilibrium, enable companies to comply with DR regulations intelligently – knowing exactly how much resourcing a specific apps needs in real-time as well as the best available location of those resources – and avoid unnecessary over-spending to maintain compliance.
Such intelligence can also help navigate longer-term issues around adapting to the new normal. Typical spikes in VDI usage once happened in the morning as users booted up, and after lunch as users came back from lunch. What some SaaS vendors – particularly those in the collaboration space – are now seeing is spikes at the top and bottom of the hour as meetings begin and end. Since traditional 9-5 office-based working hours are no longer the norm, your system needs to be intelligent to adapt not only to demand spikes caused by events, but also by the changing rhythms of a workday where the majority of employees are working the majority of their hours remotely.
AI can help overcome these overlapping and amplifying factors increasing Fiserv / Fintech IT volatility. Indeed, AI may be the only way to overcome all the tradeoffs necessary to enable a healthy equilibrium between application demand and infrastructure supply in such volatile times.
After all, consider how many tradeoffs your IT pros must consider on a daily, even hourly, basis today: around performance, at the very least they must consider overall transaction throughput and application response time; around compliance they must consider regulations on data sovereignty, affinity/anti-affinity, as well as other internal business constraints; and around cost, they must balance hardware cost and software licensing costs with budgets. And these are just the beginning – there are literally dozens of tradeoffs that IT must consider at any time to maintain this healthy equilibrium between performant applications and an infrastructure that remains in regulatory and budget compliance.
Consider, however, that 500-vm data center managing only seven tradeoffs must successfully navigate some 1.4 quadrillion combinations at any time (1,486,071,034,734,000 such combinations to be exact) to overcome typical volatility in normal times. And that a 5000-instance data center with the same seven tradeoffs must navigate 15 sextillion combinations (15,435,996,312,664,276,965,000 to be precise), one begins to realize that AI is the only answer.
Today’s volatility isn’t your father’s volatility, and it doesn’t seem to be going away anytime soon, even if this fall passes like Y2K without a hiccup.
That’s why we built Turbonomic. Come see how our AI can help you overcome volatility and adapt to our new normal.