There seem to be a fair number of generalizations and some fairly sketchy guesswork regarding how the backend infrastructure for MSFS is organized.
For example, I’ve heard a lot of people making reference to “stressed servers on azure”. If you don’t know, then PLEASE, STOP SPECULATING.
Speculative statements like these add little value to the discussion, and pollute the conversation. If you have a theory, please help out the community and test your theory. If you find something and have data to back it, make a Zendesk ticket.
Typically distributed systems like this are built upon many layers of physical and virtual infrastructure, each with a specific job or jobs, often scaling and operating independently. The systems today are vastly more sophisticated and complex than even 5 or 10 years ago. With this can come fragility.
For example, for many systems, well before an individual node is brought to its knees, most systems will try and scale, either horizontally (“more trucks”) or vertically (“bigger trucks”), or both in an effort to meet demand. Sometimes this happens “on demand”, other times scale up and/or scale out is done preemptively and kept “warm” (“at the starting line”) or “cool” (“at the paddock”). Forward of the fleet might be one or more layers of load balancers, caches, and proxies. Any one of these layers can encounter a problem that might cause downstream issues for a customer/consumer.
The truth is, we have no idea how the backend infrastructure that is responsible for map data is structured. Some or all of it may be delivered to the end user mostly from edge networks operating outside of the azure boundary. It’s also possible all the infrastructure sits inside of azure datacenters. Some of that backplane might be shared, while other parts might be dedicated.
Among the list of MANY possible problems:
- Inability to scale fast enough to meet demand
- Capacity limits on shared network infrastructure
- Traffic routing problems
- Limits on physical or virtual infrastructure
- Problems with 3rd party providers (CloudFlare, etc.)
- Software bugs, either in the game, or on the backend causing undesirable behaviours
- Problems with clocks (yes, clocks)
- A million other possible things
Bottom line, these systems can be complex, and tricky to work on at the best of times.
So if you want to help:
- Collect screenshots (the problem, and your graphics settings in game)
- Save your dxdiag
- Note where you see the problem, your altitude, and the plane you were flying
- summarize the problem and explain how to re-create the problem, and submit it to Zendesk
If you’re extra crafty you can grab debug info from the game, network traffic info, etc.