NovaStar / System Design / Overview
Introduction
A NovaStar system consists of one or more Debian Linux servers that run the NovaStar software. NovaStar servers are referred to as "nodes". The decision about how many nodes to use depends on requirements for redundancy, fail-over, on-premise or cloud hosting, etc.
For multiple-node systems, one node is designated as "primary" and receives all the data. Additional nodes are referred to as "backup" or "secondary" nodes. If the primary node has a hardware or network failure, a backup node can assume the responsibility of data collection.
The following sections describe typical system configurations. Note that even if a cloud-hosted approach is selected, some on-premise hardware components will be needed if data collection stations use local communications such as radio and ALERT/ALERT2 protocol.
Core System Design
The core system consists of data collection and integration components. The following sections describe typical configurations, in approximate order from simple to complex. In all cases:
- Appropriate access must be given to TriLynx Systems staff to provide operational support for each node, and to allow for system updates.
- Implementing public-facing websites requires additional integration, for example:
- Export data from NovaStar to other components (e.g., custom websites).
- Use data web services to access data (e.g., from Esri ArcGIS software).
- Implement a Data Node.
Single Cloud-hosted Node
- Cloud-hosted NovaStar server, typically an Amazon Web Services (EC2) instance or Google Cloud Platform Compute Engine instance.
- TriLynx can configure the node within its AWS account or work with clients to use their instance.
- Because a single node is used, there is no redundancy for failure and outages will require system support to restore functionality.
- Access to cloud-hosted machines for support is typically easier for TriLynx staff.
Single On-premise Node
- On-premise NovaStar server, which can be physical or virtual (e.g., VMware server provided by an organization).
- Because a single node is used, there is no redundancy for failure and outages will require system support to restore functionality.
Two Nodes, Both Cloud-hosted
- Because cloud-hosted virtual machines are typically very reliable, this option is not typically implemented.
Two Nodes, One On-Premise, One Cloud-hosted
- On-premise NovaStar server, which can be physical or virtual (e.g., VMware server provided by an organization).
- Cloud-hosted node provides off-site redundancy (but still may require on-premise hardware for data collection).
- Two nodes provide redundancy and fail-over to minimize impacts of local and internet outages.
- Access to cloud-hosted machines for support is typically easier for TriLynx staff.
Two Nodes, Both On-Premise
- On-premise NovaStar servers, which can be physical or virtual (e.g., VMware server provided by an organization).
- Two nodes provide redundancy and fail-over.
Data Node
It is often desirable to isolate the NovaStar data collection nodes from data publishing applications, for the following reasons:
- The local network configuration may require that the data collection system is isolated from the public internet.
- Public-facing data applications can add load if run on the data collection nodes, especially during events, when both the data collection system and data access is high.
- The use cases and users for data access may be different from the data collection system and require different optimization.
- A data publishing server can be simpler than the data collection system.
- Local add-on applications may be required, which can complicate the operations and maintenance on data collection nodes.
For the above reasons, it is recommended to implement a NovaStar "data node", which is a separate, public-facing, cloud-hosted node that focuses on publishing data. See the Data Node documentation.