Private Cloud Deployments
Every Datafye environment — whether a Foundry for algo development or a Trading Environment for live execution — runs as a private cloud deployment. This means you get a dedicated, isolated set of infrastructure resources that are yours alone.
Need help choosing? See Choose Your Path to select the scenario that fits your needs, or Choosing Your Approach for a detailed comparison. Ready to provision? Learn about The Datafye CLI and Descriptors.
What is a Private Cloud?
In the Datafye context, a "private cloud" is a completely isolated deployment of the Datafye platform components you need. Unlike shared SaaS platforms where hundreds of users compete for the same resources, each Datafye environment is dedicated exclusively to you. Your compute, memory, storage, and network resources are isolated from other users, ensuring that your data, strategies, and execution remain completely separate. This architecture provides consistent, predictable performance that isn't impacted by other users' workloads, along with security that keeps your intellectual property — your strategy logic, parameters, and data — completely private.
This level of isolation and dedicated resources ensures that whether you're backtesting a strategy or executing live trades, you get institutional-grade performance and security.
Why Private Cloud?
Performance Consistency
In shared environments, your algo might run fast during off-peak hours but slow down when other users are active. With private cloud deployments, you never have "noisy neighbors" whose workloads impact your performance. You get predictable latency for market data feeds and execution times because the CPU, memory, and network bandwidth are dedicated to your workloads alone. And when you need more resources, you can scale up without affecting or being affected by others.
This consistency is critical for algorithmic trading where milliseconds matter and reliable execution is paramount. You can't afford to have a strategy miss an opportunity because someone else's backtest is consuming shared resources.
Data Isolation
Your trading strategies represent valuable intellectual property. Private cloud deployments ensure your algo logic never leaves your environment, your historical requests and live feeds aren't shared with others, and your trades and positions remain invisible to other users. Even your configuration — parameters, credentials, and settings — stays completely isolated.
This level of isolation is standard in institutional trading environments, and Datafye brings that same protection to all users, whether you're an individual developer or a hedge fund.
Compliance and Control
Many traders and firms have compliance requirements around data handling and execution. Private clouds give you the control you need to meet those requirements. You can choose where your data and compute resources are located to satisfy data residency rules. You get complete audit trails of all activity in your environment for regulatory reporting. You control who can access your environment and can implement the access policies your compliance framework requires. And you can demonstrate to regulators that you're meeting requirements for data handling and trade execution.
Deployment Options
Datafye offers three flexible deployment models to match your needs and budget. Whether you want to start free on your laptop, leverage your existing cloud infrastructure, or have Datafye handle everything, there's a model that fits.
Standalone Deployment (Free)
All Datafye services run on a single machine. You can deploy Datafye on your own hardware — desktop, laptop, physical server, virtual server, or even a cloud IaaS instance. This option is completely free, with no Datafye infrastructure charges. You simply provide the machine and Datafye runs there using Docker containers.
This works best for individual developers learning the platform, anyone developing and testing algos before moving to production, cost-conscious users who have adequate local hardware, and anyone who wants complete control over their environment. You can provision in minutes, and you have full access to everything since it's running on your own machines.
Requirements:
Docker installed and running
8GB+ RAM recommended (16GB+ for Full Stack scenarios)
20GB+ free disk space
Stable internet connection for data feeds
Available for:
Algo development (all Foundry scenarios)
Paper trading only (not live trading)
The tradeoff is that performance is limited by your machine's resources. If you only have 8GB of RAM, you can't backtest as many symbols in parallel as you could with distributed deployments. But for development, testing new strategies on your laptop, running paper trading on a spare desktop, or learning the platform before scaling to distributed deployments, Standalone is perfect.
Note: Docker is only used for Standalone deployments. Distributed deployments use native cloud infrastructure without containerization.
Distributed Self-managed (Free)
Services are spread across multiple machines in your own cloud account. You deploy Datafye infrastructure in your own cloud provider account — AWS, Azure, GCP, or others. Datafye doesn't charge you for infrastructure; you only pay your cloud provider directly. This gives you full control since everything is deployed in your own account with complete visibility into resources and costs. You get cloud-grade scalability, able to add compute resources as needed for parallelized backtesting. And you get the low-latency data feeds and high compute capacity that cloud infrastructure provides.
Datafye deploys using native cloud infrastructure optimized for each provider. For example, on AWS, Datafye deploys directly on EC2 instances without containerization, leveraging Datafye's own deployment and monitoring capabilities rather than third-party container orchestration technologies.
This option works best for developers who want cloud scalability with direct cost visibility, teams who already have cloud accounts and infrastructure in place, users who want cloud performance without managed service fees, and organizations with existing cloud compliance frameworks they need to work within.
Requirements:
Active account with AWS, Azure, GCP, or other cloud providers
Appropriate permissions to provision resources
Datafye CLI installed and configured
Understanding of cloud resource management
Available for:
Algo development (all Foundry scenarios)
Paper trading only (not live trading)
You'll use this for production-scale algo development with cloud scalability, paper trading with institutional-grade performance, multi-strategy development needing significant compute, or when your organization has existing cloud infrastructure you want to leverage.
Distributed Datafye Managed (Paid)
Services are spread across multiple machines in a dedicated cloud account under Datafye's root account, operated by Datafye. This is a paid managed service where Datafye handles provisioning, scaling, updates, and monitoring completely. You don't need a cloud account or cloud expertise. The environment comes pre-configured for optimal performance and is production-ready from day one.
This is the right choice for users who want cloud performance without needing cloud expertise, production trading where you want Datafye to handle operations, teams who prefer managed infrastructure so they can focus on strategies, and Marketplace algos that investors deploy without needing to understand the underlying infrastructure.
Requirements:
Datafye account with billing enabled
Datafye CLI installed and configured
Payment method on file
Available for:
Algo development (all Foundry scenarios)
Paper trading
Live trading (only available with this deployment model)
Important: Live trading is only available in Distributed Datafye Managed deployments. This ensures enterprise-grade reliability, compliance, monitoring, and 24/7 operational excellence for production trading with real capital.
You'll use this for live trading with real capital, Marketplace algos deployed by investors, production environments where you want Datafye to handle all operations and monitoring, and any situation where your team wants to focus on strategies rather than infrastructure management.
Deployment Architecture
Regardless of where you deploy, each environment includes the same core components based on your scenario. The architecture scales from simple data access to full trading infrastructure:
Foundry: Data Cloud Only includes data cloud services for historical and live market data APIs, networking for your algo containers to connect (Docker networks for Standalone deployments, native cloud networking for Distributed deployments), and monitoring with health checks and status endpoints.
Foundry: Full Stack adds the algo container runtime to host your SDK-based algos, the backtesting engine for parallel backtest execution and optimization, the MCP server for AI-assisted vibe coding integration, and comprehensive monitoring and metrics for observability.
Trading: Data Cloud + Broker provides data cloud services for live market data feeds, the broker connector for integration with your brokerage account, networking for your algo containers (Docker networks for Standalone, native cloud networking for Distributed), and monitoring and alerting for trades and positions.
Trading: Full Stack combines everything: data cloud services, algo container runtime, broker connector for trade execution and portfolio management, MCP server for AI-assisted vibe coding and live iteration, and comprehensive monitoring and alerting.
Resource Isolation
Each private cloud deployment ensures complete isolation across multiple dimensions. Your compute is isolated with dedicated CPU cores for your workloads, guaranteed memory allocation, and no sharing with other users' processes, giving you predictable performance characteristics. Your network is isolated in private network segments with dedicated bandwidth, no cross-user traffic, and secure communication between components. Your storage is completely private with encryption at rest, no shared databases or filesystems, and complete data privacy. Even your data feeds are isolated with dedicated market data connections, private WebSocket streams, no sharing of live data subscriptions, and consistent feed latency.
Scaling Your Deployment
As your needs grow, you can scale your deployment in two ways. Vertical scaling adds more resources to existing components — increase CPU cores for faster backtesting, add memory for larger datasets, expand storage for more historical data, or upgrade network bandwidth for lower latency. Horizontal scaling adds more instances of components — multiple backtesting workers for parallel execution, additional data cloud instances for high-volume strategies, redundant broker connectors for failover, or distributed algo containers for multi-strategy execution.
You can also migrate between deployment models as your needs change:
Standalone → Distributed Self-managed when your strategies mature beyond single-machine resources
Distributed Self-managed → Distributed Datafye Managed when you want to transition operational responsibility to Datafye
Distributed Datafye Managed → Distributed Self-managed if you build cloud expertise and want to take back control
Datafye provides tools and support for seamless migration between deployment models.
Cost Considerations
The cost of running Datafye depends on your deployment model choice, not on which scenario you select. All four scenarios (Data Cloud Only, Full Stack Foundry, Data Cloud + Broker, Full Stack Trading) can be deployed using any of the three deployment models, with costs determined by the deployment model.
Deployment Model Costs
Standalone Deployment: Free
No Datafye infrastructure costs
You pay only for your machine (electricity for local machines, cloud VM costs if using a cloud instance)
Best for: Learning, development, and paper trading
Distributed Self-managed: Free
No Datafye infrastructure costs
You pay only your cloud provider's infrastructure costs (VMs, networking, storage)
Best for: Teams with cloud expertise who want production-grade infrastructure without managed service fees
Distributed Datafye Managed: Paid
Managed service with dedicated infrastructure operated by Datafye
Pricing includes infrastructure, monitoring, compliance, and operational support
Required for live trading
Best for: Production live trading and teams wanting Datafye to handle all operational concerns
How Scenarios Affect Costs
While costs are determined by deployment model, different scenarios have different resource requirements:
Data Cloud Only requires minimal infrastructure (just data services)
Full Stack Foundry requires more resources for backtesting and optimization engines
Trading scenarios require additional infrastructure for broker connectivity and portfolio management
Full Stack Trading requires the most comprehensive infrastructure stack
These resource differences affect your infrastructure costs in Self-managed deployments (you pay for more cloud resources) and may affect pricing in Datafye Managed deployments.
Security
All deployment types implement strong security practices. Every connection uses encryption in transit with TLS for all data feeds and API calls, and encryption at rest for stored data and credentials. Network security is enforced through firewalls and security groups protecting resources. Complete audit logging captures all system and user activity for compliance and troubleshooting.
Network Security Model
All Datafye deployments operate within private cloud environments with no public internet access by default. Datafye services run in a private Virtual Private Cloud (VPC) or equivalent isolated network. The API endpoints use fixed hostnames that vary by deployment model: localhost:8080 for Standalone deployments, api.rumi.local for internal access in Distributed deployments, or external DNS names (<user>-<type>-<env>-api.datafye.io) when external access is explicitly configured. API requests don't need authentication headers or tokens; instead, access is controlled through network security groups, firewall rules, and IP allowlisting at the infrastructure layer. Private connectivity is typically provided through site-to-site VPN, AWS PrivateLink, Azure Private Link, or GCP Private Service Connect. See the API Reference for complete details on API URLs and access models.
Best Practices
Even within a private cloud, follow these security best practices to maintain a secure environment:
Minimize Network Access — Restrict API access to only the containers and services that need it
Use Internal DNS — Configure private DNS zones for Datafye services
Monitor Access — Enable network flow logs to track API access patterns
Secure Your Containers — Follow container security best practices for your own algo containers
Credential Management — Use secrets management services (AWS Secrets Manager, Azure Key Vault, etc.) for broker and data provider credentials
For cloud deployments, whether in your account or Datafye-managed, additional security layers include cloud provider security groups and network ACLs, IAM-based access control, VPC isolation, DDoS protection, and automated security patches.
Monitoring and Observability
Every private cloud deployment includes monitoring capabilities to help you understand what's happening in your environment. Health checks verify all components are running correctly. Performance metrics track CPU, memory, network, and latency. Algo metrics capture strategy performance, signals generated, and positions held. Error tracking provides centralized logging of errors and exceptions. And alerting sends notifications for critical issues that need attention.
For Datafye-managed deployments, Datafye's operations team monitors your environment 24/7 and handles issues proactively before they impact your trading.
High Availability
Production deployments can be configured for high availability to ensure your trading continues even when individual components fail. This includes redundant components that provide failover for critical services, automatic restart so services recover from failures without manual intervention, data replication for backup and disaster recovery, and geographic distribution with multi-region deployments for resilience against regional outages.
High availability configurations are available for cloud deployments, whether in your account or Datafye-managed. Local deployments are inherently single-node, so high availability requires cloud deployment.
Getting Started
Ready to provision your first private cloud deployment? Here's the path forward:
Choose your path — Review Choose Your Path to decide which scenario fits your needs
Create your descriptors — Review Descriptors to create configuration files for your environment
Install the CLI — Follow CLI Installation to get the Datafye CLI
Understand the CLI — Learn how the Datafye CLI uses descriptors to provision deployments
Provision your environment — Use the appropriate provision command for your scenario:
foundry provision for Foundry environments
trading provision for Trading environments
Next Steps
Understand the platform — Read What is Datafye? to learn about the platform's architecture and capabilities
Understand deployment scenarios — Read Choosing Your Approach to understand which scenario fits your use case
Learn about the CLI — Read The Datafye CLI to understand how to provision and manage deployments
Learn about descriptors — Review Descriptors for an overview, then dive into Data Descriptors, Algo Descriptors, and Broker Descriptors to understand how to configure your deployment declaratively
Explore AI-assisted development — Learn about The Datafye MCP Servers for natural language algo development
Last updated