Cloud-native systems replace servers with services and that makes all the difference, says Olga Orekhvo, COO at CompatibL, via an FTF News Q&A.
(Over the past two years, securities firms have begun to embrace the next chapter of cloud computing — cloud-native systems that exploit services and pipelines. This is essential for firms that want to reap the full benefits of the cloud, says Olga Orekhvo, chief operating officer (COO) at CompatibL Technologies. FTF News recently got time with Orekhvo, which resulted in a two-part Q&A series. This second part of the interview focuses on why the cloud-native approach has so much appeal to financial services firms.)
Q: Could you provide your definition of what a cloud-native system is?
A: A cloud-native system is a software application that runs on a private or public cloud, uses a services-based architecture that enables massive scalability, and takes advantage of development pipelines.
The first two of these criteria are mentioned frequently, but the third is not always on the list.
At CompatibL, we believe that development pipelines — automated processes for performing application build, automated testing, configuration, deployment, and scaling — are indispensable to realizing the benefits of the cloud and speeding up the development, test, and production phases for each software release.
When it comes to non-cloud-native software, you have to manually install it on one or multiple servers. All processing is done on the servers you configured, and you have to manually scale it up when needed.
The scalability of computing, database, IO, etc. depends on the ability to provision new servers and reconfiguring the application. And if these servers are down for maintenance or as a result of a breakdown, the application has to be switched to a different set of servers.
In a cloud-native system, servers are replaced by services.
Unlike servers, services are virtualized, distributed, and scalable, and their deployment is standards-based and is managed through the cloud using a separate set of deployment services.
There is no longer a specific set of servers to deploy the application to, and the whole system is not dependent on an individual machine that may malfunction at any time. Relying on services for all aspects of the application — from its core code to object storage, messaging, and databases — strengthen a company’s resilience to data loss and make scaling much easier.
But taking a legacy application and merely moving it to virtualized cloud instances along with all its supporting apps/services (cache, database, queue, scheduling, etc.) does not make it cloud-native, even though it’s technically running in the cloud.
To be cloud-native, the application needs to use services and pipelines. Only then can the full benefits of the cloud be realized.
To harness the benefits of cloud-native architecture while providing our clients with maximum flexibility of deployment options, we designed CompatibL Risk Cloud as a multi-cloud solution that can run on virtually any set of cloud technologies on any cloud provider.
The cloud-native architecture of the platform makes it possible to deploy the same version of the CompatibL software on any private, public, or hybrid cloud, using the client’s preferred container orchestration, storage, messaging, and caching solutions. This possibility to use the client’s preferred cloud provider and technologies eliminates the need for a protracted procurement process and security reviews and ensures a fast and effective deployment process.
Q: What are the benefits of a cloud-native solution compared to a system that was moved to a cloud?
A: When faced with a complex project with tough deadlines and budget limitations, a project’s leadership often takes the path of least resistance and simply transplants the existing software architecture to the cloud by relying on the Infrastructure-as-a-Service (IaaS) offering of their cloud provider.
This option results in migrating the application by merely creating its mirror image in the cloud, with each on-premises server migrated to its virtualized cloud counterpart.
While this approach requires few changes in the software itself, eventually it deprives organizations of the tremendous value and potential of cloud-native technologies.
To unlock the full potential of the cloud, the software must be re-engineered around services and deployment pipelines. It must use modern, cloud-native technologies such as serverless computing, distributed NoSQL [structured query language] databases, cloud object storage, and others.
Olga Orekhvo
Only by fully embracing cloud-native architecture, using services-based technologies and tools, and demanding the same of the firm’s vendors can a financial services firm realize the full benefits of the cloud.
Among the many advantages of running cloud-native systems is enhanced cybersecurity.
A cloud-native application reduces cybersecurity risks by securing a standard set of cloud services and technologies, which present less penetration risk than non-standard, on-premises or hybrid networks. This presents a significantly reduced risk of a data breach compared to relying on protection of custom instances.
Another benefit of the cloud is its high degree of redundancy even when running in a single geographic region. Even higher redundancy can be achieved by using a different geographic region as a hot standby, this leads to significant cost savings compared to deploying an on-premises application at a disaster recovery site.
Cloud-native solutions can take full advantage of the cloud services “pay-per-use” billing model, which helps optimize the cost of variable application loads.
Financial services firms run many compute-intensive processes as part of their end-of-day workflows as well as intraday, for example during portfolio optimization or when executing a new trade.
Using legacy architecture on-premises or in the cloud for variable loads results in ineffective resource use as either a physical or virtualized instance cannot be ramped up or down seamlessly when the application is running; the ability to ramp up or down a cloud-native service is readily available.
In one of our case studies, by implementing serverless cloud-native technologies, we were able to reduce the infrastructure cost of CompatibL Risk Cloud deployment on AWS [Amazon Web Services] by an average of 62 percent, and on [Microsoft] Azure by an average of 72 percent compared with deploying the same application in an on-premises data center.
This three- to four-fold cost advantage was made possible by leveraging true serverless cloud-native technologies rather than running the existing application on virtualized servers.
Q: Are there cloud-native innovations on the way for the CompatibL Cloud offering? If so, what are they and when will they become available?
A: Machine learning is rapidly becoming a must-have in financial trading and risk applications. At CompatibL, we are at the forefront of this change.
CompatibL is conducting cutting-edge research on machine learning models for portfolio valuation and risk management and was the first vendor to deploy a machine learning credit limit model in production.
As part of the next release of our platform planned for Q1 of 2022, we plan to release a full suite or machine learning models for valuation and risk across all asset classes.
Because of the cloud’s ability to rapidly harness the massive computing resources that are required for training a neural network, successful production deployments of machine learning algorithms are cloud-native. They rely on the cloud’s pay per use model to optimize the cost by ramping up the resources only when they are needed – for training machine learning models.
After the training is completed, resources can be ramped down to lower levels that are appropriate for using the models that are already trained. For this reason, CompatibL will always be relying on cloud-native approach deploying our machine learning models to production.
Need a Reprint?