Hi,
There isn't really a universal sizing calculator for Parallels RAS, as the required resources depend heavily on the number of concurrent users, their workloads, applications, and infrastructure architecture.
However, Parallels and Microsoft provide some useful official sizing guidelines that can help you get started.
Parallels RAS hardware requirements:
https://kb.parallels.com/en/128884
This article provides some useful recommendations:
- Light workloads: up to 6 users per vCPU
- Medium workloads: up to 4 users per vCPU
- Heavy workloads: up to 2 users per vCPU
- Power users: up to 1 user per vCPU
- RAS Secure Gateway: recommended maximum of 1,200 users per server in Gateway SSL mode
- For smaller environments (up to approximately 30 users), a single-server deployment may be possible.
These figures are indicative and should not be treated as guaranteed capacities.
Microsoft also provides detailed RDS session host sizing guidelines:
https://learn.microsoft.com/en-us/w...ession-host-virtual-machine-sizing-guidelines
Microsoft recommends considering:
- Concurrent users rather than total users
- CPU and RAM consumption per workload
- Storage performance and profile management
- Logon storms and peak activity
- Virtualization overhead
- High availability and failover capacity
For HALB sizing, Parallels has published additional recommendations:
https://kb.parallels.com/en/123606/
For example, the updated HALB recommendations include configurations ranging from 2 vCPUs / 2 GB RAM for up to 12,500 concurrent TCP sessions under low-intensity traffic, to 8 vCPUs / 4 GB RAM for up to 20,000 TCP sessions under moderate-intensity traffic.
Please note that TCP sessions are not equivalent to concurrent RAS users.
There is also an older Parallels HALB scalability test documenting a recommended limit of 2,000 user sessions per appliance:
https://kb.parallels.com/en/125229
Ultimately, I would recommend starting with the expected number of concurrent users and their application workloads, then validating the sizing through a pilot deployment with real performance monitoring.
For larger production environments requiring high availability, a commonly used architecture includes:
- 3 RAS Connection Brokers
- 2 RAS Secure Gateways
- 2 HALB appliances configured for high availability
This provides redundancy across the main RAS infrastructure components and helps avoid single points of failure.
Of course, the actual number of components depends on the environment size, expected workload, network topology, and availability requirements.
This does not include optional or additional components such as Enrollment Servers, Reporting, and Monitoring, which may also need to be considered depending on the deployment.
It's worth noting that the number of Connection Brokers, Secure Gateways, and HALB appliances should be determined by both availability requirements and actual capacity needs, rather than simply by the number of users.
Hope this helps!
Thierry