General System Description

System Function or Purpose is 18F’s product line for the tools, tech, and services 18F provides to help teams delivering federal digital services to operate efficiently at-scale in a cloud-hosted environment, while complying with federal regulatory requirements. It’s based on and built using the open source Cloud Foundry project, which is an open platform as a service, providing a choice of developer frameworks, and application services which makes it faster and easier to build, test, deploy, and scale applications. is a self-contained platform as a service (PaaS) which can be used on top of underlying infrastructures such as AWS, VMware and Open Stack.

Information System Components and Boundaries

18F has created a specific set of VPCs (Live production, tooling, and staging) for All VPCs have subnets used to separate and control IP address space within each individual VPC. Subnets must be created in order to launch Availability Zone (AZ) specific services within a VPC. 18F has setup VPC Peering between the Staging VPC, tooling VPC and the CF Live production VPC.

Cloud Foundry components include a self-service application execution engine, an automation engine for application deployment and lifecycle management, and a scriptable command line interface (CLI), as well as integration with development tools to ease deployment processes. It has an open architecture that includes a buildpack mechanism for adding frameworks, and application services interface, and a cloud provider interface.

The PaaS Information System is hosted within the AWS East Public Cloud in the Northern Virginia Region. AWS services utilized include ENI, EC2, EBS, VPC, RDS, S3, MFA, Route 53, ELB and IAM. These are listed as leveraged hardware, network and server components.

Physical aspects of the PasS information system are outside of the authorization boundary due to all hardware being physically managed by AWS. While other services are reviewed and approved for use by the GSA OCISO, they were deemed to be ancillary support services that do not directly process/store data but rather provide general support services. These services include Cloudwatch, Cloudtrail, CloudFormation, AWS Config and Trusted Advisor.

The authorization boundary diagram represented within Figure 9-1 depicts the core components which make up the System in its entirety. The diagram specifically depicts, outlined in red, those components that are within the authorization boundary and those that are outside of the boundary. AWS components outside the boundary belong to an authorized System thus allowing for inheritance of the component. Other support services outside the information system authorization boundary include, Pager Duty, New Relic, Slack, Trello, GitHub, Code Climate and Cloudability. Pagerduty, Slack and Trello are communication tools used by 18F developers and support staff while New Relic, GitHub and Code climate are developer based tools in support of the system. .

Network Architecture

The following architectural diagram(s) provides a visual depiction of the system network components that constitute Security Domain Stack

Identification and Authentication Control

18F staff utilizes multi-factor authentication (MFA) for AWS IAM, GitHub, Trello and Slack. Access to these services requires multi-factor authentication using both the login and the MFA device. When a user signs into these services, they are prompted for an authentication code from their MFA device. Once the user has entered their username and password along with their authentication code, they will then be granted access. Slack, GitHub and Trello use authentication codes delivered to users’ cell phones generated by an application such as Google authenticator or sent as a text message (SMS). Trello also performs pass-thru authentication by using google authenticator or GSA’s SecureAuth and a GSA email account of the user.

AWS Identity and Access management (IAM) is used to manage 18F users and their privileges that use the virtual private cloud environment. Each user is assigned a specific role or multiple roles within IAM to ensure they are only allowed authorized privileges and functions required to perform their job. All access to IAM requires multi-factor authentication through AWS MFA. 18F staff uses a mobile application such as google authenticator to receive randomly generated authorization codes to access the AWS management console and services.

Client users log into the web portal with a username and password. All traffic from the public internet to the Cloud Controller API end point and UAA happens over HTTPS. Its primary role is as an OAuth2 provider, issuing tokens for client applications to use when they act on behalf of users. 18F is in the process of implementing a multi-factor solution for client agencies to leverage. It’s currently a responsibility of client agencies to enable multi-factor authentication for its users when required for its applications and to platform. Once users have authenticated to, clients are able to see the Organizations and spaces that belong to them. The following table lists the URLs associated with platform as a service. The and URLs are out of scope as they are not associates specifically for the platform.

Site Name URL Function Main Landing Page Client user login screen Same as login screen Same as login screen Handles all API calls to provides status of services

Designated 18F DevOps users who require access to the underlying platform login to the jump box using SSH credentials. MFA functionality for the jump box is on 18F’s roadmap to implement for privileged access. The jump box is a virtual machine (VM) that acts as an access point for the BOSH Director and deployed VMs. Direct access to other VMs is disabled. The jump box is located in the us-east-1d Availability zone. Privileged access to the jump box is restricted to originate only from GSA IP space and a limited set of IPs that 18F DevOps use to access from the office.

ACLs, Software defined Firewalls and Security Groups

The VPC is an underlying control domain that provides a layer of protection to enhance each boundary, both external and internal. The VPC is configured to utilize routing tables, network access control lists, subnet rules, and security group (firewall) rules. Each of these controls must have appropriate rules and routes in-place before any external service is able to reach a host within

Amazon S3 Access Control Lists (ACLs) enable 18F to manage access to buckets and objects. Each bucket and object has an ACL attached to it as a sub-resource. It defines which AWS accounts or groups are granted access and the type of access. When a request is received against a resource, Amazon S3 checks the corresponding ACL to verify the requester has the necessary access permissions. When a bucket or an object is created, Amazon S3 creates a default ACL that grants the resource owner full control over the resource. A grantee can be an AWS account or one of the predefined Amazon S3 groups.

Each EC2 instance or group of instances is assigned to at least one Security Group with specific assigned roles. These roles delineate what traffic is allowed inbound & outbound for each instance by allowing or denying ports, protocols, and services. Access controls within EC2 include, secure login to instances using key pairs and username, a firewall that enables the specification of ports, protocols and source IP ranges that can reach instances.

Cloud Foundry application security groups (ASGs) control the traffic flowing out of applications. Each cf application uses a dedicated Linux container, and each container includes a dedicated virtual network interface. Application controls are defined by a CF administrator, and the application specific policies are stored and then applied to the virtual network interface by the platform before an application even starts. Application security groups are a collection of ‘allow’ rules that can be made with global or application specific assignments enabling access to be set on individual application requirements. These requirements are added through whitelisting and whitelisting is layered on top of a series of container-centric lock-downs, allowing limited access to other applications and services.

The platform is further secured using defined route tables. Route tables contain a set of rules, called routes, which are used to determine where network traffic is directed. Each subnet within a VPC is associated with a route table; the table controls the routing for the subnet. A subnet can only be associated with one route table at a time, but you can associate multiple subnets with the same route table. For a subnet to be accessible from the internet, it needs to be associated with a routing table and have an internet gateway as the default gateway.

Audit Logging, Monitoring and intrusion detection utilizes the Elasticsearch Logstash & Kibana (ELK) Stack for centralized audit logging and monitoring of components, applications, and data APIs. The logs captured from the ELK stack can be used in forensic analysis to track down the time of intrusion, as well as the method used to penetrate into the network. It’s used in a way to monitor attacker activities in the network and help determine the reason behind an attack.

The ELK stack includes Logstash, a centralized logging and parsing data pipeline that is used to process logs in different formats. Logstash uses different rules to format each log message into multiple fields, which are indexed by the Elasticsearch search engine used for deep searches and data analytics. Kibana is a web interface that provides an overview of the collected data, so 18F can easily view and analyze the collected logs. With the Grafana add on, 18F has a centralized dashboard for all events and metrics monitored within its environment.

Suricata is a high performance Network IDS/IPS and Network Security Monitoring engine. It implements a complete signature language to match known threats, policy violations and malicious behavior. It detects many anomalies in the traffic it inspects and is capable of using the specialized Emerging Threats Suricata rule set and the VRT rule set. It uses JSON events and alert output which allows for easy integration with Logstash. 18F plans to implement Suricata with the ELK stack. utilizes CloudTrail and CloudWatch for all logging and monitoring of its virtual cloud infrastructure and components. 18F has implemented CloudTrail for its account monitoring. It provides visibility into user activity by recording API calls made on an AWS account. CloudTrail records important information about each API call, including the name of the API, the identity of the caller, the time of the API call, the request parameters, and the response elements returned by the AWS service. This information helps 18F track changes made to its AWS resources and to troubleshoot operational issues.

18F has implemented CloudTrail for its account monitoring. It provides visibility into user activity by recording API calls made on an AWS account. CloudTrail records important information about each API call, including the name of the API, the identity of the caller, the time of the API call, the request parameters, and the response elements returned by the AWS service. This information helps 18F track changes made to its AWS resources and to troubleshoot operational issues.

18F has implemented CloudWatch for its resource monitoring. It allows 18F to monitor AWS resources in near real-time, including Amazon EC2 instances, Amazon EBS volumes, Elastic Load Balancers, and Amazon RDS DB instances. Metrics such as CPU utilization, latency, and request counts are provided automatically for these AWS resources. It allows 18F to supply logs or custom application and system metrics, such as memory usage, transaction volumes, or error rates. All VPC Flow logs, CloudWatch and CloudTrail data in relation to are consumed by the ELK stack and visualized through a Grafana dashboard.

Vulnerability Scanning, Penetration testing

18F utilizes Nessus for vulnerability scanning and asset management of its network environment. 18F conducts baseline configurations scans, compliance scans, virtual infrastructure and network scans using custom scan policies and templates. The Nessus manager is located in a separate VPC, which is a peer of Live VPC. Public access to the Nessus Manager is limited within the GSA IP address only. It runs both authenticated/unauthenticated scans against 18F assets and provides reporting and grouping functions of vulnerabilities with scanned assets. Tenable Nessus allows 18F to track risk level changes based on remediation efforts and provides a variety of scan report outputs. Nessus reports 18F assets vulnerable to common exploits. Metasploit is a penetration testing platform designed to exploit and validate vulnerabilities found within 18F’s virtual private cloud and applications.

18F utilizes OWASP ZAP for internal web application scanning of information systems and components. It is used to scan for the OWASP Top 10 vulnerabilities within applications and it can perform enhanced automated penetration testing functions.

Cloud Inventory and Asset Management

18F utilizes VisualOps Cloud management to provide a visual, real-time representation of 18F’s virtual cloud infrastructure. It also provides a global view of the 18F AWS account where all regions and services can be seen in one place. Visual Ops has a WYSIWYG editor to design, configure and provision AWS cloud applications. Once applications are deployed, VisualOps monitors and manages the apps to ensure they are running in the defined states.

AWS Config can export a complete inventory of AWS resources with all configuration details, determine how a resource was configured at any point in time, and get notified via Amazon SNS when the configuration of a resource changes. AWS Config can provide configuration snapshots, which is a point-in-time capture of all 18F resources and their configurations. Configuration snapshots are generated on demand via the AWS CLI, or API, and delivered to an Amazon S3 bucket that is specified.

18F uses Cloudability for AWS cost and usage management of the AWS virtual infrastructure. It provides graphical information of detailed billing of the AWS account information through analytics on usage and consumption across 18F’s AWS account.

Static and Dynamic Code Analysis

Code Analysis for is performed by Code Climate which is integrated by Cloud Foundry in GitHub to perform all static code analysis for the platform. Code Climate is an automated static code analysis for the platform as a service that is It integrates directly into 18F's continuous integration (CI) workflow and provides notifications of actionable findings in GitHub, and Slack when an issue arises. 18F uses a variety of other analyzers based on the type of codebase used on top of the platform. OWASP Zap is used as an dynamic code analysis tool. It has the capability to dynamically probe web applications looking for unknown vulnerabilities found in the source code.

Incident Response Resolution and Communication

18F utilizes PagerDuty for Incident response and communication. It integrates with 18Fs entire monitoring and deployment stack which includes Slack, Cloudwatch, New Relic and ELK with real-time alerts and visibility into critical systems and applications. It allows the DevOps team to quickly detect, triage, and resolve incidents from development through production.

The SecOps team can view centralized alerts from ELK and CloudCheckr for 18F’s visibility of the entire application stack and virtual infrastructure which enables the team to prioritize, escalate and execute the incident response process.

Configuration Management and Version Control

18F’s configuration management process consists of GitHub, BOSH and Concourse for Baselines and configuration files are stored in GitHub repositories using version control. These files are then deployed using a pipeline yaml file within the Concourse continuous integration platform. Concourse does not allow any configuration to enter the server that cannot be easily saved away in version control. Builds run inside their own containers so that installing packages on the build machine doesn't pollute other builds.

CloudFormation templates are utilized to deploy the AWS infrastructure stack within the VPC. Configuration management at the AWS level is managed through CloudFormation templates that describe in JSON (JavaScript Object Notation) format the entire set of resources to be created by AWS. EC2 instances, S3 buckets, VPCs, subnets and security groups are resources that can be deployed by CloudFormation. AWS Cloudformation templates help maintain a strict configuration management scheme of the cloud infrastructure.

AWS Config can export a complete inventory of AWS resources with all configuration details, determine how a resource was configured at any point in time, and get notified via Amazon SNS when the configuration of a resource changes. AWS Config can provide configuration snapshots, which is a point-in-time capture of all 18F resources and their configurations. Configuration snapshots are generated on demand via the AWS CLI, or API, and delivered to an S3 bucket that is specified.

System Environment

The following describes the Platform as a service (PaaS) as deployed in production based on the network diagram above. The Cloud Foundry platform as a service is implemented as hardened EC2 instances or virtual servers operating within Cloud Foundry Virtual Private Cloud (VPC). The instances are deployed to grow and shrink based on demand.

To provide high availability,'s blob store is on AWS S3, the backend DB is on AWS Relational Database Service (RDS), which is hosted in two different geographically separate locations and the load is distributed amongst the instances (runners) running cloud foundry. The backups of the database are configured with a Retention Point Objective (RPO) within the last five minutes. is a system with multiple components. Only "runner" instances have containers. Traffic goes from the elastic load balancer (ELB) through the "router" to the "runner". When an application is deployed to, an image is created for it and stored internally. The image is then deployed to a Warden container to run in. For multiple instances, multiple images are started on multiple containers. Cloud Foundry's internal Controller leverages the underlying infrastructure to spin up virtual machines to run the Warden containers. When an application is terminated, all its VMs can be recycled for another application to use. If the application instance crashes, its container is killed and a new Warden container is started automatically. A container only ever runs one application ensuring isolation, security and resilience.

A load-balancing router sits at the front of to route incoming requests to the correct application - essentially to one of the containers where the application is running. The only access points visible on a public network are a load balancer that maps to one or more routers and, optionally, a NAT VM and a Bastion Host (jump box). Because of the limited number of contact points with the public internet, the surface area for possible security vulnerabilities is minimized.

Applications deployed to access external resources via Services. Applications can also access any service that can be reached through the network (either public or private). In the environment, all external dependencies such as databases, messaging systems, and file systems are considered Services. When an application is pushed to Cloud Foundry, the services it should consumed also can be specified. Services have to be deployed to the platform first and then are available to any application using it. Virtual Private Cloud (VPC) environment

Public Subnet

The hosts within the Public subnet are only used for system maintenance and DNS purposes only. Instances do not contain any GSA operational data. The Public subnet allows only the necessary inbound and outbound access with the Internet defined by AWS Security Groups. In the public subnet there exists a Cloud Foundry bastion host (jump box), A VM that acts as a single access point for the deployed instances as disabling direct access to the instances is a common security measure. The bastion host (jump box) is used for the purpose of executing the BOSH commands within Cloud Foundry. ACL’s for the in scope resources can he reviewed at the following location: ACL’s

Private Subnet - Core Tier

The Core Tier is a production subnet where the instances reside. Access to this subnet is restricted to only administrators. All application development, maintenance and deployment are conducted within the Core Tier subnet.

Private Subnet - Services Tier

The Services Tier is a production subnet where applications deployed to Cloud Foundry within the Core Tier can access external resources via Services maintained in the Services Tier through the private network. In the environment, all external dependencies such as databases (i.e., Mongo DB, PostgreSQL) and messaging systems are within the Services Tier.

Private Subnet - Database Tier

The Database Tier hosts the configurations within a PostgreSQL database (i.e. users, orgs, and spaces). These are hosted in geographically separated, isolated by failure physical locations; in short the RDS database is in a Multi-AZ deployment.

Private Subnet - Metrics

The Metrics subnet includes’s continuous integration continuous deployment (CI/CD) pipeline. Baseline and configuration files are stored in GitHub repositories using version control. These files are then deployed using a pipeline yaml file within the Concourse continuous integration platform. It also hosts the ELK Stack logging and collection tools used for PaaS internal system environment

Cloud Foundry has a flexible organizational structure representation defined through Organizations, Spaces, and Roles.

Organizations “Orgs”

At the top of the structural hierarchy are Organizations. They serve as the boundary for governance provided to all the users in that Organization along three different dimensions: quotas, service availability, and custom domains and routes. This allows a single instance of Cloud Foundry to host different Organizations representing different companies, departments, or even projects, and keep separate the resources associated with each of them. Organizations can be defined any way that you like. An org contains at least one space and can have multiple spaces within it. All collaborators can access an org with user accounts.


Within organizations are sets of spaces. Spaces define a more narrowed set of resources, scoped to a particular project, team, or environment. All application development and maintenance has to happen inside of a space. Spaces are where developers, QA, and operations usually are found. A space gives access to a shared location for application development, deployment, and maintenance, and users will have specific space-related roles.

User Accounts

All users have their employee status categorized with a sensitivity level in accordance with PS-2. Employees (or contractors) of service providers are considered Internal Users. All other users are considered External Users. User privileges (authorization permission after authentication takes place) are described in the table that follows. At this time all users who have access to the 18F AWS VPC and the underlying PaaS are internal users. There is no external access granted to users outside of the 18F/USDS programs and GSA.

A user account represents an individual person within the context of a Cloud Foundry installation. A user can have different roles in different spaces within an org, governing what level and type of access they have within that space. The combination of these roles defines the user’s overall permissions in the org and within specific spaces in that org. A list of standard cloud foundry user account types can be found in the table below.

Types of Users

Role Internal or External Sensitivity Level Authorized Privileges and Functions Performed
Org Manager Internal High Add and manage users, View users and edit org roles, View the org quota, Create, view, edit, and delete spaces, Invite and manage users in spaces, View the status, number of instances, service bindings, and resource use of each application in every space in the org, Add domains
Org Auditor Internal Low View users and org roles, View the org quota
Space Manager Internal High Add and manage users in the space, View the status, number of instances, service bindings, and resource use of each application in the space
Space Developer Internal Moderate Deploy an application, Start or stop an application, Rename an application, Delete an application, Create, view, edit, and delete services in a space, Bind or unbind a service to an application, Rename a space, View the status, number of instances, service bindings, and resource use of each application in the space, Change the number of instances, memory allocation, and disk limit of each application in the space, Associate an internal or external URL with an application
Space Auditor Internal Low View the status, number of instances, service bindings, and resource use of each application in the space

Hardware Inventory

None - Leveraged from AWS Infrastructure

Software Inventory

Hostname Function Version Patch Level Virtual (Yes / No)
Cloud Controller (CC) Primary API entry point for Cloud Foundry. It’s responsible for managing the lifecycle of applications. The Cloud Controller also maintains records of orgs, spaces, services, service instances, user roles, and more Version 2 212 Yes
GOrouter Central router that manages traffic to applications deployed on Cloud Foundry Version 2 212 Yes
Droplet execution agent (DEA) Performs staging and hosting applications Version 2 212 Yes
Health manager monitors the state of the applications and ensures that started applications are indeed running, their versions and number of instances correct Version 2 212 Yes
User Account and Authentication (UAA) identity management service for Cloud Foundry Version 2 212 Yes
Login Server Handles authentication for Cloud Foundry and delegates all other identity management tasks to the UAA. Also provides OAuth2 endpoints issuing tokens to client apps for Cloud Foundry (the tokens come from the UAA and no data are stored locally). Version 2 212 Yes
Collector discover the various components on the message bus and query their /healthz and /varz interfaces Version 2 212 Yes
Loggregator user application logging subsystem for Cloud Foundry Version 2 212 Yes
Jump Box Used to run BOSH commands Version 2 212 Yes
BOSH open source tool chain for release engineering, deployment and lifecycle management of large scale distributed services Version 2 152 Yes
ELK Logging Front end for consuming logging and monitoring data for metrics reporting Version 2 2.2 Yes
Concourse CI Performs continuous deployment and integration of components Version 0.75 0.75 Yes
Grafana Centralized dashboard for all events and metrics for Version 2 212 Yes
Code Climate CloudFoundry Static code analysis tool for the source code Latest Version NA Yes
GitHub CloudFondry and code repository and version control Latest Version NA Yes AMI operating system used in deployed EC2 instances for Ubuntu 12.04 LTS 12 Yes

The following table lists all other applications and components used in relation to the PaaS information system.

Component Name Function Version Patch Level Virtual
Nessus Vulnerability scanner for the platform Latest Version NA Yes
New Relic Application and performance metrics for Latest Version NA Yes
Cloudability Monitor, optimize and govern Cloud Costs related to the AWS Infrastructure Latest Version NA Yes
Slack Message, alert and communications app Latest Version NA Yes

Trello| Used for agile, scrum and story boarding| Latest Version | NA | Yes| |VisualOps| Visual DevOps tool for AWS| Latest Version| NA | Yes| |Pager Duty| Incident communication platform used by DevOps| Latest Version|

Network Inventory

None - Leveraged from AWS Infrastructure

Ports, Protocols and Services

Ports (T or U) Protocols Services Purpose Used By
443 (T) HTTPS HTTPS CF ec2 web service AWS, Cloud Foundry
80 (T) HTTP HTTP CF ec2 web service AWS, Cloud Foundry
22 (T) SSH Secure Shell (SSH) Secure command line interface AWS, Cloud Foundry Jumpbox
53 (U) DNS DNS Service Inbound DNS requests AWS, Cloud Foundry
4222 (T) TCP NATS bus messaging service Provides publish-subscribe and distributed queueing messaging system internally between all CF components Cloud Foundry
6868 (T) HTTP Bosh agent interface The Agent executes tasks in response to messages it receives from the Director. Cloud Foundry
123 (T) NTP Network Time Protocol (NTP) Sync time within the network Cloud Foundry, AWS CloudTrail, Syslogs
17 UDP QOTD Provide quote of the day
1 ICMP Internet Control Message Information and diagnostics for network devices (Ping) AWS, Cloud Foundry
6 (T) TCP AWS Elastic Cache Mem Cache AWS, Cloud Foundry
2222 (T) SSH Secure Shell deamon (SSH) External port for SSH access for apps CF SSH Jump box
4443 (T) TCP WSS TCP /Websocket Traffic Cloud Foundry
5432 TCP PostgreSQL Director bosh_db CF BOSH
8081 (T) TCP User account and authentication server (UAA) Provide identity management and authorization services CF UAA, Login Server
25250 (T) TCP BOSH Blobstore repository where BOSH stores release artifacts, logs, stemcells, and other content, at various times during the lifecycle of a BOSH release Cloud Foundry Jumpbox
25555 (T) HTTP BOSH Director Coordinates the Agents and responds to user requests and system events. Cloud Foundry Jumpbox
25777 (T) TCP BOSH BOSH Registry Cloud Foundry Jumpbox

System Interconnections

None - Not Applicable

results matching ""

    No results matching ""