Planned Nimbleness: A Practical Guide for Asset Managers

Broad acceptance of cloud-based platforms in asset management has provided a wealth of opportunities for improved security, scalability, reliability, and performance. While cloud has also lowered barriers to entry for startup managers, pitfalls loom large and need to be kept in mind when architecting a reliable and nimble system. 

Nimbleness, for asset managers, is a competitive advantage. The ability to test investment ideas quickly and comprehensively, to implement changes to an investment process easily, and to monitor efficacy and improvements allow the investment team to put their best thinking to work confidently without unnecessary delay. 

The idea of ‘planned nimbleness’ refers to how your infrastructure and software are architected to allow quick, robust changes to your system with high confidence and low risk. One important dimension of infrastructure nimbleness relates to scalability. This is primarily what is provided by a cloud deployment. 

A helpful attribute of cloud products is that they provide an interactive console that gives access to all services available. While this is an easy way to explore and experiment with what’s available, using the console as a primary interface to stand up and shut down resources can quickly become tedious. Configuring through the console is time-consuming and error-prone because of its repetitive nature. Standardized configuration norms have to be entered each time to go through a workflow. 

A remedy to this challenge comes in the form of cloud service’s ability to script resource management in a way that makes it much less error-prone and easier to enforce configuration defaults. There are third-party products that allow an architect to generalize an infrastructure across vendors, which helps avoid vendor lock-in. These scripts can be used to stand up the infrastructure in a different cloud region—or even a different platform—for testing, BCP, or migration.

This kind of infrastructure-as-a-service (IaaS) scripting gives the flexibility for the codification of infrastructure in a way that is executable, versioned and easily tested. While the console is a useful tool for getting familiar with a service and to better understand its capabilities, it’s important to move its deployment to a scripted implementation as soon as possible.

Considering architecture design

Security is a critical concern in any cloud deployment, and it’s important to understand the implications of network architecture design. The default settings provided when creating private clouds, subnets, and compute instances may not provide the level of protection that an organization needs. Therefore, it’s important to decide on those configuration values and use the IaaS scripting functionality to standardize the creation of network enabled components in an environment. Some areas to focus on include public IP addresses, VPN, segregation of duties (development vs. test vs. production), and access to the internet.

Development environments like Jupyter Notebooks and scripting languages like Python have significantly reduced the barriers to entry to programming. This is useful for those building tools and algorithms for investment research and portfolio management. However, good software architecture becomes even more important as a result of having more programmers of varying skill levels in the mix. 
 
In the same way that IaaS scripting helps the architect take a step back from using the interactive console as the primary interface, architecting a software library to perform common tasks can alleviate the risks of developers re-inventing the wheel whenever they need to perform a common task. 

Investment professionals vary in their programming skills, so it’s not surprising to see beginner programmers working alongside programmers who have been coding for decades. By abstracting away low-level functions, a common library allows developers to focus on what’s important to the investment business logic. Automating library packaging and deployment makes it easy to do this. It provides a mechanism to standardize activities—such as file access, logging, database connectivity and credentials management—across the organization. 

What works well is to start small and design the libraries to be easily extended and redeployed. This is where automation and testing are important. The architect should continually review the functionality built by developers and identify what should be promoted in the library.

Working out the costs

Closely related to software architecture is the process of continuous integration—the idea of taking every commit to the source code repository and sending it through a pipeline of checks and tests, before allowing the code to be made available in production. These checks include style guidelines, formatting, type checking (for scripting languages), and, most importantly, test cases. Any failure in these steps could halt deployment. 

Continuous integration/continuous deployment (CI/CD) in cloud platforms are typically serverless, making it simple to control costs. Almost all programming languages have a test framework that easily plugs into these CI/CD pipelines. Test cases are key to enabling organizations to quickly push changes through to production. 

Organizations are exploring the move to the cloud for many reasons, but cost is the most common among them. On the surface and from a cost perspective, the idea of pay-as-you-go seems easier compared to building and managing a data center. Organizations can pay for services by the hour or gigabyte, but be warned: costs can spiral out of control without constant monitoring. Organizations can use scripting to deploy resources, as it’s simple to build scripts to ensure compliance with expectations.  

Serverless architectures (containers, lambda) are another way to perform regular activities without deploying compute instances. The organization pays only for compute time and can avoid costs related to idle time.

Using the cloud to be more nimble

The financial services industry is considered a late adopter in its move to the cloud, largely due to security concerns. As a result of advances to security in the cloud, issues related to multi-tenancy and public access to resources have been greatly mitigated. As with any tool, it is possible to configure a system to bypass default configurations meant to protect data, so it’s important to be aware of how data and resources could be made publicly available. 

Organizations should continually review the security they have in place and take advantage of security surveys that the cloud provider offers. They should also consider using multi-factor authentication (MFA), single-sign-on (SSO) and the principle of least access.

While starting from scratch makes it easier to implement these best practices, it’s possible to take an existing deployment and migrate to a more nimble architecture piece by piece. The flexibility and scalability of the cloud allow organizations to start small and thoroughly test along the way at little cost.

As cloud service offerings continue to grow quickly, it’s important to stay current regarding products and service enhancements, which can often simplify workflows. It also pays to continue to monitor the deployment to avoid unexpected costs or inefficiencies.

Nimbleness is a key advantage to investment firms because it allows quick and reliable deployment of new ideas. Architecting the environment using the practical ideas outlined above will make a team more efficient and confident as changes are promoted to production.

Mauricio Bustos is a partner and head of data and technology at RadiantESG Global Investors.