Return of the composable enterprise
Gartner, in its latest Press Release (August, 2020) has identified "Composite Architecture" as one of the five emerging trends that will drive technology innovation for the next 5 to 10 years. Composite Architecture, which creates the foundation for a Composable Enterprise, is agile by design with the ability to respond to the changing business environment rapidly. According to Gartner, "By 2023, 60% of mainstream organizations will list composable enterprise as a strategic objective and will use an increasing number of packaged business capabilities."
Let me go back by 8-10 years to call out what the thought leaders were saying at that point of time :
In April 2013, Jonathan Murray, coined the term in his blog "The Composable Enterprise". He argued, than in order for an enterprise to be hyper-competitive in the fast changing digital world, it has to move from a static and functionally siloed Operating Model to a model where the various business capabilities can be conceived as lego-blocks which are loosely coupled and can be assembled to configure new services and re-configured to meet the needs of the changing competitive landscape. In such a scenario, IT needs to be much more dynamically adaptable in order to respond to the needs of business in a matter of hours and days rather than months and years. The Composable Enterprise therefore leads to the Composable Operating Model which in turn needs a Composable IT Architecture (often called Composable Architecture Model or CAM), the key attributes of which are elastic infrastructure, auto provisioning, unified master data model, integrated analytics, atomic services, anytime-anywhere-any device access.
A year later, McKinsey in its paper "A two-speed IT Architecture for the digital enterprise", recommended a fast-speed, customer-centric front end running alongside a slow-speed, transaction-focused legacy back end. This essentially meant that, the customer facing systems are the real digital systems (channels) which are "any time, anywhere and any device" and imparts the lego-block capability to the digital stack of applications. Add, remove, reconfigure any time with little or no operational impact while you can keep your legacy applications intact at the bottom of the stack.
The third point is around a Research Report from Forrester (maybe around 2011) which predicted that by 2020, the role of the CIO's will change from “owners and operators of technology components” to “architects & brokers” of technology & business components. They called the regime as Empowered Business Technology(BT).
The common thread of these 3 propositions is an architectural re-imagination which is fundamental to digital re-imagination especially for the enterprises who are not born in the digital age and are not digital native. Therefore one can argue whether digital transformation without addressing the underlying enterprise architecture in a systematic manner is an effective approach at all.
Here are some of my views as underlying tenets of the Composable Architecture :
1. Microservices and Containerization by far are the most important architectural construct for a Composable Enterprise. It is like the glue which binds the lego blocks of capabilities together and is therefore already in the mind map of most CIO's. Not only this helps to expose atomic and self contained business services but also it can wrap around one or multiple services exposed by back end system/ external partners etc., with necessary orchestration. These microservices, exposed as API's are the building blocks of the composable enterprise.
2. What to do with the legacy core transactional systems ? This is probably the most important dilemma as there is an inertia among CIO's to hold on and protect the investments made in the legacy and avoid the complexity of change management of a big bang or multi wave replacement. Here, the goal will be to do away with the "Data Gravity" and "Entanglement" by breaking the monolith and opt for Packaged Business Capabilities (PBC) wherever possible. Some of these PBC today are coming from domain specific & niche SaaS platforms who aim to solve a particular business problem in a particular domain while some can also be developed as self contained micro services. In case, the legacy ERP is too customized and entangled, then the best approach would be to go for the 2 speed architecture, but wrap the capabilities with Microservices before it is consumed by the front end systems of engagements. This Gartner blog, beautifully explains the future of ERP (legacy is here to stay !) and literally echoes the 2-speed architecture adoption for a composable enterprise.
3. Unified Data Model - Establishing a single source of truth across business for all kinds of master data be it product, services, customers etc.
4. Data & Intelligence has become ubiquitous, real time and on demand. Datawarehouses are out and Data Platforms are in. Given the high degree of skills and complexity (Data Ingestion, Pipelines, Data Lakes/ Ponds, Data Quality, Governance, Privacy & Consent, Data Processing Infrastructure, Data Ops, ML Model training infrastructure, Model Deployment & Management - MLOps etc.) that are needed to set up, run and manage a Data Platform, I foresee most CIO's will source their needs from 3rd party specialized data platform providers who will ingest the enterprise data and provide api's for the enterprises to consume the intelligence coming out of it. Also the future Datwarehouses will probably be built on top of a data fabric based on Knowledge Graph which will stitch different types of data along with their semantic relationship and leverage the in memory graph processing technologies and the native graph algorithms to generate insights.
5. A high speed data bus will be necessary to exchange data fragments between each block, in real time and with the ability to trigger events. Looking at the business requirements around customer experience, real time and event based is now a mandate for going digital.
6. At the experience layer, frameworks like Domain Driven Design & its variant the Event Storming Framework will be used to quickly design immersive experiences, often leveraging VR/AR/MR technologies and consuming the business capabilities and intelligence exposed by the microservices. Business users should be able to create front end applications on a self service mode and access data on demand.
7. Security & data privacy will be pervasive and immersive across all the components.
8. Finally, the IT will have to increasingly play the role of architects and brokers of these components. The "caretaker" and "manager" of IT assets approach and mindset will have to be done away with. Here Gartner talks about "fusion teams". If we look at the event storming approach, its a mixed team of Business, IT, UI/UX experts who collaborates to create the journeys.
The existing Enterprise Architecture, as we see today, was laid out at a time to support static and analog business models. As enterprises become more digitally enabled and the business model becomes dynamic, there is little doubt that future applications will be built using composable blocks of business capabilities that can rapidly enable new business processes. It is definitely a a significant architectural shift, but a much needed one !
***
Disclosure : The author is an employee of Group Digital Office, Tata Sons. The opinions expressed herein are his own and do not reflect those of the company.
I like the idea very much, however, in IT it isn't easy. The idea of the 'composable enterprise' sounds pretty much like Mr Warnecke gras formulated it in his book 'The Fractal Company: A Revolution in Corporate Culture', published 1993.
Dear Sarthak, well articulated article. Agree completely on common services data model. At Servicenow we are delivering great value to our customers by using our Common Services Data Model Data Base and building entity relationships to proactively service a client, employee, device or a business. It's a platform with packaged apps and Lego block building capability to quickly release apps by using No Code Low Code platform capabilities. Microservices on legacy apps is also true can be plugged into the agile bus to move information, tasks and accountability like blood to every one needy round the clock in a scalable non stop Cloud.
I see a maturity in ESB for EAI or A2A services while microservices for the users - systems interactions and the IoT (device sensors - systems interactions) - the technicalities will be maintained - it means - queueing - streaming - replications - RFC - http and traditional ftp - while optimising the movement of (either data closer to the execution node or execution closer to the data node)
Interesting read..The challenge I see is in the ERP SAP kind of system wherein creating Microservices on top of it looks overkill ..not sure how they react..SAP might have to create more block within itself..