Ah... Computer hardware! To many an unknown subject, sometimes scary, or in some cases familiar. Author of original illustration: Nocturnalfrog, modified for Alt Under Ctrl by Wrox.
Server Infrastructure & Management
The Alt Under Ctrl crew has an assigned member who manages the server infrastructure as their primary task, which currently is Wrox. Alt Under Ctrl (from now on "the community") owns or rents a total of 3 dedicated servers responsible for web pages, the forum and databases to store both web and game data, and the in-game playgrounds such as game servers.Resource usage
A majority of the community's infrastructure is hosted on virtual machines (from now on "VM:s"), which are managed by the dedicated servers (referred to as "Hypervisors" in the Status & Data page). This method of using resources effectively splits processing power and memory within a dedicated server to several smaller servers, the VM:s. Additionally, this adds an extra layer of management. So why the usage of VM:s?
A VM is in the depicted hypervisor illustration one of several
VM:s hosted within one dedicated server.
Why virtual machines?
As a majority of tasks and feature additions on the servers hosting the game servers are performed by humans, potentially fatal mistakes as a result of humanity can lead to simultaneous downtime of several services, such as web, database, game and content (fast download) servers, if those are not separated.This can contribute to a fall in availability as services depending on databases, for example game servers with the "Bunny Hop" game mode would cease to respond. At the same time, the forum would not load for visitors. In addition, the content server serving files to gamers connecting to the community's servers would reject requests, effectively rendering new players unable to connect to any servers with custom maps, as they would not be able to download them.
The community's strive is to reduce the likelihood of this type of larger disruption from occurring. Splitting resources into their own instances is a step to increase redundancy and stability. In the event of one VM failing, the rest remain available. However, hypervisors hosting VM:s require a certain level of resources to avoid performance degradation while operating virtual machines. What hardware has the community decided to utilize for this task?
Hypervisor Hardware
| Location | CPU (L3 Cache, GHz Turbo, Cores/Threads, Benchmark Score*) | Memory (DDR Type, (ECC), Speed) | Motherboard |
|---|---|---|---|
| Chicago, IL, United States | Intel® Xeon® E3-1270 v3 (8M Cache, 3.90 GHz, 4 C/8 T ~9900) | Samsung 32GB DDR3 1600 MHz | SuperMicro X10SLD-F |
| London, United Kingdom | Intel® Xeon® E3-1270 v5 (8M Cache, 4.00 GHz, 4 C/8 T ~10000) | Micron 32GB DDR4 2400 MHz | SuperMicro X11SSE-F |
| Stockholm, Sweden | 2x Intel® Xeon® E5-2670 (40M Cache, 3.30 GHz, 16 C/32 T, ~19000) | Hynix 48GB DDR3 ECC 1333 MHz | ASUS Z9PA-D8 |
Locations that have been discontinued
The following locations have been discontinued to effectivize and refocus services to other near and equivalent or better locations in terms of available resources:- Helsinki, Finland
- Roubaix, France
- Amsterdam, Netherlands
- Naaldwijk, Netherlands
- Oslo, Norway
- Dublin, Ireland
- Karlskrona, Sweden
- Bristol, United Kingdom
- Atlanta, GA, United States
- New York City, NY, United States
- San Jose City, CA, United States
- Los Angeles, CA, United States
Active game servers previously hosted in the specified locations have been moved to a near location.
