Skip to main content

Physical Hosting

Introduction​

Physical hosting involves running DHIS2 servers within a country, where organizations manage their own physical servers either within ministry premises or centrally in a national data center. In cases where the ministry or government has sufficient technical and human resources and infrastructure, physical hosting can be a preferred option.

As mentioned, physical hosting can further be sub-divided into local hosting and centralized hosting.

Local Physical hosting​

The Ministry of Health typically installs its server in a dedicated room. This server room often houses multiple servers, not just one. These servers are often clustered together to ensure high availability and reliability.

Advantages​

  • Direct Control: Offers the highest level of control over server resources and data.
  • Accessibility: Having the server on-site ensures easy physical access for maintenance,

Drawbacks​

  • Demands significant technical expertise for design, installation, and maintenance.
  • Initial investment costs are high.
  • Presents various physical challenges such as power supply reliability, pest control, environmental control (water, ventilation), physical security, and round-the-clock building access.

Because of the costs and risk involved, in most cases we would not advise going this route.

Centralized physical hosting​

The Ministry of Health (MoH) applications reside in a central government data center, acting as a shared service for various ministries and sectors. This data center typically uses virtualization technology to create and share resources efficiently.

In simpler terms, MoH is just one tenant in this large data center "cloud" environment, sharing resources with other government entities.

Advantages​

  • Aligned with national policies on data sovereignty
  • MoH does not need to maintain server infrastructure directly
  • Only System Administrator skills are required

Drawbacks​

  • Cost may be higher than cloud hosting providers
  • Dependent on the skill level and infrastructure of the data center team
  • Access to services can be hampered by bureaucratic processes
  • We often see performance issues in these setups

Other Considerations for Physical hosting​

The server​

It is the server where you would typically have compute and memory resources. More often that not, you'd also have directly attached storage on a physical server. A data-center in most cases will have more than one server running, and different servers serve different purposes depending on the application they are running. Workloads have different resource requirements, there are those that are compute intensive, others are over reliant on memory and so forth. Servers are typically interconnected using fast switches.

A PostgreSQL server thrives on good resources, including a good share of CPU and RAM. For optimal performance, fast disks are essential, with Solid State Drives (SSDs) being the preferred choice. The following command, for example, provides a quick overview of your disk performance in terms of latency. Any latency exceeding 3 seconds may indicate a potential issue.

dd if=/dev/zero of=/tmp/test2.img bs=512 count=1000 oflag=dsync

Networking​

In cases where you are building a cluster of servers, you will typically need to have a fast connectivity between your physical servers. It is recommended to have fiber links supporting even up to 40G links connecting your servers. This implies a modern network Switch in your server room. If you are going to be running your workloads on a Network Attached Storage (NAS), ensure this resource has excellent networking to your server stack. Clustered hosts will typically need super-fast networking to perform optimally.

Internet Connectivity​

Inbound Traffic: DHIS2 application access for users will occur through the internet. This necessitates sufficient bandwidth to accommodate user traffic efficiently.

Outbound Traffic: Periodic software and system updates require outbound internet access for the server.

Recommendation: To ensure smooth user experience and efficient update processes, a high-bandwidth internet connection (e.g. 1 Gigabit per second) is recommended for your data center.

DMZ network​

In a Data-Center, you have a gateway to your infrastructure, this is usually a reverse proxy. The gateway needs to have public ip address for it to be accessible from the internet. It is a good practice to have this server in a separate network, a DMZ. Another server that can be run in this network is an SSH JumpHost.

Internal network​

You have other servers that should not to be accessible from the internet. Examples are the application server, this is were you'll be running your dhis2 application, and the database server, this is where you'll have PostgreSQL database running. These servers should be on a private network with no direct inbound access from the internet. These servers are then only accessible from either ssh JumpHost for ssh access for from the proxy (running nginx, apache2 or HAProxy) for web app access.

High Availability​

A high availability (HA) setup is broadly defined as infrastructure without a single point of failure. Its a good practice to have redundancy on the different levels i.e server, networking and storage.

Backups and disaster planning​

Backups and archiving requires additional server resources. Consider a disaster recovery plan with backups in a separate data center for added security.

Testing environment​

A dedicated testing environment (separate servers) is crucial for safely testing DHIS2 updates and other major changes before deploying them to production. This minimizes risk to your live system.

Budget for additional servers to facilitate a robust testing environment. Plan for major upgrades to operating systems and database servers every 2-3 years.

Testing these upgrades in the dedicated environment helps ensure a smooth transition when implemented.

Prioritize Physical Security:​

Implement Rigorous Access Control: Enforce strict access protocols for the data center. This may include the use of security badges, biometric authentication systems, and security cameras to monitor entry points and critical areas.

Mitigate Environmental Hazards: Develop a comprehensive pest control program to safeguard against rodents and other potential threats. Additionally, consider measures to mitigate other environmental hazards like flooding or fire.

Maintain Cable Infrastructure: Regularly inspect and maintain network and power cables to prevent damage or deterioration.