Arun Janarthnam, George Panitsas, Gopinath Durairaj, George Tsolis, and Jungang Wei contributed to this blog post.

Well-architected solutions are truly something special. They enable us to focus on what we need to do without worrying about underlying technical details. To us as users, they just work.

You can think of Citrix Analytics like that. Every day, customers around the globe depend on it to generate key insights for their digital workspace. It helps answer such fundamental questions as, “Which users are experiencing poor session performance, and which are a higher security risk to the organization?”

We do this by processing vast amounts of data with cloud technologies and machine learning. Customers don’t need to worry about all the backend “magic” that Citrix Analytics makes happen. It just works.

In this blog post, I’ll share some architectural details and stats about the Citrix Analytics platform to give readers a sense for how we’ve architected it as enterprise-grade at cloud-scale.

Unified Platform for Data, ML, and Analytics Apps

Citrix Analytics is a three-tiered platform. At the lowest level, the Data Platform ingests data events across a wide array of Citrix and non-Citrix products and technologies. These data are cleansed and securely stored in the data lake for consumption by the machine learning (ML) layer. The ML Platform is where the computational modeling is done for classifications of performance or risk conditions, prescriptive recommendations, and other functions. Cloud-native technologies with sheer scale and processing power make this possible. It’s one of the key reasons why a solution like Citrix Analytics can only live in the cloud.

Of course, you can connect both on-premises and cloud-based Citrix environments to the platform. This happens through our next tier — the Application Platform. This is where data sources are defined, UI elements live, searches and alerts take place, and much more. All this is visualized in a simplified diagram.

Data Platform

Data is one of Citrix’s most important assets, and the Data Platform is responsible for collecting, processing, analyzing, and presenting that data. It’s a centralized hub for all Citrix data and breaks individual data “silos” by ingesting data across various organizations to a centralized location and transforming it into a common format. This common format enables Citrix engineers to obtain better insights by enriching datasets with other internal or external datasets.

The Data Platform democratizes data by providing a secure and easy way for authorized Citrix employees to access and query data. By democratizing data, we empower product teams to convert their ideas into successful products and cloud services that help Citrix’s customers.

The Data Platform’s real-time data processing engine enables us to react to security threats and other events in under a minute. In addition to these core capabilities, the Data Platform also has out-of-the-box support for scaling, data governance, data accuracy and security. The Data Platform is the foundational fabric on which the ML Platform and Application Platform are built.

ML Platform

The ML Platform supports data scientists, data engineers, and ML engineers throughout the stages of research, development, and operationalization of ML solutions. Developers have access to a rich collection of datasets through a notebook-based environment, equipped with tools well established in the ML/AI community, streamlining data exploration, model experimentation, and product development. ML-based solution operationalization is backed by a distributed processing framework, which is enriched with configuration, testing, scheduling, and monitoring capabilities, forming a clean and efficient development SDK.

The selected technology stack supports demanding ML solutions, with increased model accuracy driven by periodic model trainings over big data, with models that can be trained/tuned on a per-customer basis but are deployed on a multi-tenant platform (supporting multiple products and multiple customers). The ML Platform is architected with both the data scientists and Citrix customers in mind, guaranteeing sustainable ML value in Citrix products.

Application Platform

Once data are transformed and insights are generated, they are presented to our customers via APIs and apps. The Application Platform supports product teams to bring these apps and APIs to market faster by providing containers to run those apps (Kubernetes), a standardized way of programmatically accessing data (GraphQL), and a micro-frontend framework to support visualizations.

The Application Platform also enables users to listen for certain events and take actions on them. For example, a customer can watch out for a user logging in from an unusual location and set a policy to log off and restrict further access to that user until further review. This simple yet powerful feature is just one example of the Application Platform’s capabilities. Availability of an app to its users is a very critical requirement. The Application Platform promises 99.95 percent uptime by providing out-of-the-box monitoring, canary, and active-active deployment support.

Reliability and Resiliency

While outages of analytics platforms, like Citrix Analytics, won’t typically affect the operation of any underlying data source, that doesn’t mean that we haven’t architected Citrix Analytics with the highest possible levels of reliability and resiliency. Once customers start using Citrix Analytics, they quickly realize how valuable the platform is to securing and optimizing their digital workspace.

The image below shows some of the key capabilities we’ve employed to this extent like full redundancy with no single point of failure and modern, container-based frontends serving up purpose-built microservices. Plus, our canary release strategy ensures that new feature rollouts occur in thoughtful waves with the ability to quickly rollback if needed.

Capacity and Scalability

Many of our customers have very large, globally distributed Citrix environments, some with tens of thousands of VDI hosts, and an equally large user base. To accommodate the sheer volume of data these environments generate (along with many other smaller environments), we’ve leveraged several cloud-native Azure services in Citrix Analytics to ensure seamless operations today and in the years to come. By the numbers, we are currently:

  • Ingesting tens of thousands of events per second into the platform
  • Processing terabytes of data every day, with pipeline latency measured in seconds
  • Managing petabytes of data in object-based data lake storage
  • Serving thousands of analytical queries every day, with sub-second response times

Data Governance and Security

Data governance and security is a topic Citrix takes very seriously. Customers are trusting us with one of their most valuable assets — their data. To ensure compliance and high security standards, Citrix Analytics:

  • Requires all traffic be secured with SSL
  • Encrypts data-at-rest using 256-bit AES encryption
  • Uses 512-bit, regularly regenerated keys for storage account access
  • Runs anonymized datasets for internal analytics
  • Enforces RBAC and change control management

Citrix Analytics has also obtained industry certifications such as ISO 27001, SOC2, and HIPAA and it adheres to GDPR guidelines. You can learn more at www.citrix.com/trust.

Uncover Your Own Insights

We want Citrix Analytics to operate as seamlessly as possible for our customers, and that’s how we’ve architected it. You just set it up and it just works.

We’ve covered several key topics here on how we designed and implemented the solution, but there’s so much more to Citrix Analytics. We encourage you to try it out for yourself if you haven’t already. Kick off a demo or run a free trial from Citrix Cloud today. See what insights and recommendations Citrix Analytics can deliver in your specific environment while taking a more proactive approach. Got a question or comment? Just leave it below and we’ll look it over.