As I meet with our customers, there are always a range of discussions regarding the use of the cloud for financial services data and analytics. Customers vary widely on the topic of public cloud – what data sources, what use cases are right for public cloud deployments – beyond sandbox, experimentation efforts. Private cloud continues to gain traction with firms realizing the benefits of greater flexibility and dynamic scalability. Then there are the more extensive discussions – scrutiny of the overarching, data strategy questions related to privacy, security, data governance /access and regulatory oversight. These are not straightforward decisions, especially when data breaches always hit the top of the news headlines. However, one strategy is consistently discussed and deployed – a hybrid data cloud. A hybrid approach offers the most flexibility and the confidence and agility to quickly pivot your strategy as needed. As firms more and more align towards a hybrid data cloud to balance their business needs, there is another layer of decisioning – private cloud, public cloud, multi-cloud.
Let’s start with our definition of a hybrid data cloud.
- Simple, consistent and intuitive experience for data users and developers
- Common tools for management, observability, performance and costs
- Security and governance consistent across distributed clouds
- Data services for structured, unstructured, real-time & batch data across clouds
- Portable, interoperable data services for the lifecycle of data across clouds
- Separated compute and storage for scale and agility
- Orchestration and management of automated workflows
- A decision framework to automate and optimize workload execution
- Open and extensible to support new clouds, data types and data services
All in a distributed cloud model that spans multi-public, private & edge clouds.
Here, I’ll provide some guidance on key considerations for a hybrid data cloud. It is not intended to be an exhaustive list, rather some of the issues that rise to the top of the strategy when assessing your hybrid data cloud deployment.
Focus on Business Strategy First
At Cloudera, we firmly advocate that before you start assessing data infrastructure deployments, you must understand the purpose of the data. What are your business goals, what are you trying to achieve? Are you seeking to improve the speed of regulatory reporting, enhance credit decisioning, personalize the customer journey, reduce false positives, reduce data warehouse costs? Answers to these questions will guide the next set of questions…
What data do I need to achieve these objectives? The question of the data to use will include the basics of transactional and enterprise data sources, but should expand to broader questions that will further shape the deployment strategy including third-party data, the need for real-time and/or unstructured data, ML and AI tools, etc.
Finally, this will lead to the operational aspects of the data usage and choice of deployments including considerations related to who needs access to the data with the associated data governance and operational reliability.
Hybrid Data Cloud includes a Multi-cloud approach
Across my discussions with customers and prospects, a hybrid environment is the dominant model present or evolving in financial services firms. Yet, the hybrid profile varies from firm to firm. The chart below from our recent Enterprise Data Maturity Research Report is not exclusive to financial services, but highlights this trend across enterprises from a range of industries.
A key observation that stands out to me from the chart is the expectation for greater multi-cloud deployments going forward, specifically 36% in 18 months’ time. This aligns with my discussions about deployment choices, specifically related to flexibility, cost management and regulatory oversight prevalent for financial services.
Our customers recognize the benefits of a dynamically scalable environment to reduce costs. Axis Bank and Kasikorn bank are examples of customers successfully utilizing the flexibility of cloud to expand and contract processing power as needed. This way they only pay for what they need. The cloud is perfect for workloads with uneven compute requirements.
The further advancement of this using multiple clouds is the ability to choose the most cost effective cloud service provider as needed. An open, multi-cloud architecture offers the flexibility to choose workload locations. I see this option becoming more prevalent as an effective cost management approach, distributing workloads based on competitive pricing models.
Managing Cloud Concentration Risk
Cloud Concentration Risk is another topic that is raising in priority somewhat depending on your locale currently that can benefit from a multi-cloud approach. The UK’s Prudential Regulation Authority (PRA) recently highlighted concerns over the systemic risk based on a concentration in cloud service providers. DORA, the EU’s Digital Operations Resilience Act includes as part of its directive a goal to de-risk third-party concentration risk. In the US, the Financial Stability Oversight Council (FSOC) is under continuous pressure to provide greater oversight on cloud service providers in order to protect against financial instability.
With this regulation circling, the multi-cloud hybrid approach can offer benefits, whether planned directly or indirectly. A multi-cloud approach enables an individual organization to be nimble in preparation for legislation as it unfolds. Our previous cloud concentration risk research highlights many of the benefits to individual organizations as well as the financial services sector as a whole of improved collaboration across cloud service providers, regulators and financial services entities.
Who Needs the Data
Another critical consideration for the hybrid data cloud is who requires access to the data. More and more I hear about the lines of business people seeking additional intelligence and access to the data. I say “people” intentionally, not “users”. We all know data is power; data is information. More and more business people want the access to that data to slice and dice as they have business ideas and assumptions that they want to explore. My observation is that we need to anticipate more and more the people that want access to data but are not data scientists or SQL experts. They need access and tools to help them. We’ve seen this already of course, and should expect more of it. We need to make the data available to these people and help them to interpret it.
As a result, we need to allow for the proper data access and controls in the context of the hybrid cloud environment. This means all of the data governance, security and privacy considerations have to be at the core of the data strategy, not an after-thought. We must address retaining data context, lineage and accurate audit trails in the highly sensitive world of finance.
Cloudera’s Shared Data Experience, SDX addresses these data control concerns for hybrid data cloud environments. SDX delivers an integrated set of security and governance across public and private clouds.
The Future Is Hybrid… as defined by Your Firm
Whether your data is currently on-premise, in public or private cloud, I expect it is in a constant state of evaluation as more demands are placed on the data, digital transformation initiatives evolve, and more PEOPLE request access. As AI initiatives advance and regulations unfold, the deployment strategy for data will need to advance as well. Creating a blueprint for a hybrid data cloud is the first step towards creating a nimble, enterprise data strategy that is proactive and responsive to industry developments and can position your organization for success over the long term.