Skip to main content
Skip to footer
Contact Us
We are taking you to another website now.
September 28, 2020

That every company is a software company is no overstatement in today’s world. According to a recent report(1), an average enterprise deploys 464 custom applications; for larger ones (>50k employees) the number stands at 788. However, such pervasiveness comes at a cost. Another recent survey found that a single hour of downtime now costs 98% of firms at least USD 100,000. The increasing size and complexity of digital infrastructure due to AI based components and hybrid deployments across cloud and on-premise infrastructure pose a major challenge in mitigating their disruption. 

Modern business applications are expected to support required throughput, latency and fulfil other SLA requirements like reliability and robustness, in addition to their core functionality. At the scale of enterprises, such non-functional requirements pose a larger challenge for enterprise IT systems.

Distributed deployment at scale and agile software development expose applications to various factors that can cause performance disruptions like unavailability and high latencies. External factors such as increase in number of users, types of users, and increase in data size; and internal factors such as failure of hardware, software bug, memory leak, and synchronization should be factored in.

There is an underlying need for a deeper adaptation of enterprise IT systems and proactive maintenance processes in the modern business environment. The systems should be able to flag possible failures and risk forecasts at the earliest so that corrective measures can be taken well before disruption strikes.  

The Solution: Performance Digital Twin

Digital twin (a digital replica of a system, product, process, or service) is a powerful idea employed for large physical systems like boilers, engines, IoT systems and even enterprises. In fact, “Digital Twin” has  featured in Gartner’s Strategic Technology trends for the last four consecutive years (2016-2020)(2)

A digital twin essentially models the behavior of the corresponding physical entity and continuously learns by consuming data (e.g. sensor readings, load, etc.) from multiple sources to stay updated and accurate.

At the enterprise level, digital twin technology helps emulate the business operating model. A digital twin for an enterprise IT system will comprise a set of AI-powered digital twins for individual components of the architecture, such as the transaction processing system, database system, and network system. They can provide accurate insights into the functioning and performance guarantees of respective components. Enterprise digital twins play a major role in identifying the bottlenecks in current business processes, driving business expansions, increasing customer base, streamlining customer journeys, and improving overall customer satisfaction. For instance, in ecommerce, proactively identifying potential hiccups such as slow response time and mitigating it using a performance digital twin can significantly benefit business.

A digital twin may also be used to understand the behavior of an enterprise IT system due to expansions and other unforeseen situations. Sometimes, existing deployment may fail to accommodate new business requirements. In such cases, making proactive migrations, scale-up/scale-out enables administrators to drive proactive maintenance of performance and ensure seamless functioning of the enterprise IT system. 

Ideally, there should be a two-way feedback loop between the enterprise IT system and its corresponding digital twin. This will enable “what-if” and “if-what” kind of experimentation scenarios for different deployment stacks and execution infrastructure. Imagine the idea of experimenting with an application which is in production (changing load, deployment platform, etc.), using its performance modeling digital twin without introducing actual downtime. This is analogous to experiencing aviation in a flight simulator. 

Such powerful simulations supported by the digital twin technology can enable enterprises to foresee deployment issues that would not otherwise have been visible at that point. A performance digital twin not only mitigates the issues during actual deployment, but also reduces the turnaround time for adapting to new or changing business requirements through proactive maintenance. 

Ideally, a performance digital twin should also have a digital twin for modeling the customer journey (i.e., customer behavior model).

Employ digital twin technology early on for best results

Co-designing the application and its digital twin leveraging agile frameworks helps save effort and time as data collected during development and testing can be used to train the digital twin. 

Just as enterprises today deploy data lakes to consolidate and analyze their data and gain business insights, they also need to deploy digital twins across their applications to mitigate business downtime.

An AI-powered digital twin of enterprise IT systems can enable seamless customer journeys to accelerate digital transformation of enterprises



[2] Digital Twins in Gartner Technology Trends

Dr. Rekha Singhal is a senior research scientist with TCS Research and Innovation.  She has worked in both corporate and academic research. Her research interests are in the areas of high performance data analytics systems, heterogeneous architecture, performance modelling of big data systems, query optimization, and storage area networks. Rekha headed the development of the disaster recovery product Revival 1000, which was launched at the Centre for Development of Advanced Computing. She received her MTech and PhD (computer science) from IIT Delhi, India. 

Follow Rekha Singhal on social media


Dr. Mayank Mishra is part of "Computing Systems Research Group" at TCS. His research group focusses on High Performance Computing (HPC) involving both software and hardware systems. He is involved in building frameworks and platforms for high performance Machine Learning and Deep Learning applications.  

Follow Dr. Mayank Mishra on social media

Dr. Shruti Kunde is part of "Computing Systems Research Group" at TCS. Her research group focusses on High Performance Computing (HPC) involving both software and hardware systems. She is involved in building frameworks and platforms for high performance Machine Learning and Deep Learning applications.

Follow Dr. Shruti Kunde on LinkedIn


Thank you for downloading

Your opinion counts! Let us know what you think by choosing one option below.