Let’s face it. Server infrastructure can make or break your multiplayer game. Without reliable gameplay experiences, players will go elsewhere. That’s where Amazon GameLift comes into play. Dedicated game server hosting in the cloud means minimizing latency and wait time, and maximizing cost savings. The difference is in the numbers.
Part of a series on |
Internet hosting service |
---|
Full-featured hosting |
Web hosting |
Application-specific web hosting |
|
By content format |
Other types |
A game server (also sometimes referred to as a host) is a server which is the authoritative source of events in a multiplayer video game. The server transmits enough data about its internal state to allow its connected clients to maintain their own accurate version of the game world for display to players. They also receive and process each player's input.
- 1Types
- 1.2Listen server
Types[edit]
Dedicated server[edit]
Dedicated servers simulate game worlds without supporting direct input or output, except that required for their administration. Players must connect to the server with separate client programs in order to see and interact with the game.[1][2]
The foremost advantage of dedicated servers is their suitability for hosting in professional data centers, with all of the reliability and performance benefits that entails. Remote hosting also eliminates the low-latency advantage that would otherwise be held by any player who hosts and connects to a server from the same machine or local network.[1]
Dedicated servers cost money to run, however. Cost are sometimes met by a game's developers (particularly on consoles) and sometimes by clan groups, but in either case, the public is reliant on third parties providing servers to connect to. For this reason, most games which use dedicated servers also provide listen server support.[1] Players of these games will often times host servers for the public and their clans, either by hosting a server instance from their own hardware, or by renting from a game server hosting provider.
Listen server[edit]
Listen servers run in the same process as a game client. They otherwise function like dedicated servers, but typically have the disadvantage of having to communicate with remote players over the residential internet connection of the hosting player. Performance is also reduced by the simple fact that the machine running the server is also generating an output image. Furthermore, listen servers grant anyone playing on them directly a large latency advantage over other players and cease to exist when that player leaves the game.[1][3]
However, listen servers have the advantage of being essentially free and not requiring any special infrastructure or forward planning to set up, which makes them common at LAN parties where latency and bandwidth issues are not a concern. They are also common in console games.
Host Migration[edit]
In a listen server arrangement, 'host migration' is a useful feature. Without host migration, if the player that is currently hosting disconnects for any reason (quitting, crashing, lost network connection, etc), the current server stops functioning and gameplay ends. A host migration feature allows one of the other players to become designated as the new host, so that the game can continue.
Peer-to-Peer[edit]
In the client/server model outlined elsewhere in this article, clients receive processed data from the server and display it without much thought. In the alternative 'peer-to-peer' model there is no server: each 'peer' instead receives the raw input streams of each other player and determines the results itself.[4]
Peer-to-peer is generally considered obsolete for action games, but it still common in the real-time strategy genre due to its suitability for games with large numbers of tokens and small numbers of players. Instead of constantly transmitting the positions of 1000 troops, the game can make a one-off transmission of the fact that 1000 troops are selected and that the player in command of them just issued a move order.[4]
However, peer-to-peer has many disadvantages:[4]
- It is very difficult to keep all peers synchronised. Minute differences between peers can escalate over time to game-breaking paradoxes.
- It is very difficult to support new peers joining part-way through a game.
- Each peer must communicate with all other peers, limiting the number of connected players.[3]
- Each peer must wait for every other peer's message before simulating the next 'network frame', resulting in all players experiencing the same latency as the player with the worst connection
Listen-peer[edit]
Multiple listen servers collectively peer amongst themselves for listen-peer setting. This type of setting avoids the disadvantages of peer-to-peer communications among all clients and is a better alternative to dedicated servers for increased number of clients.
Tickrate[edit]
The rate at which a game server runs simulation steps is commonly referred to as its 'tickrate'. A 'tick' is a number associated with each simulation step which is broadcast to clients to help them synchronise with the server.[5]
There are three reasons to limit the frequency of server simulation steps to a predefined tickrate: to conserve server and client bandwidth, to conserve server CPU time, and to allow clients to be certain of how much time has elapsed between each tick. The last point is important for internet games, as network updates from the server can arrive at different intervals or even an incorrect order.[5]
Customization[edit]
Servers, particularly those of PC games, can generally be customized in ways that still allow unaltered clients to connect to them. These customizations can include tweaks to built-in game settings, content that is downloaded by clients when they join the game,[6] and new code which changes the way that the server behaves.[7]
While server customization is popular with server administrators and players, it can be at odds with the desire of developers and other players for the game to be experienced as intended.[8][9] It can also aggravate players by enabling abusive administrators to lie about what their server offers.[10]
See also[edit]
References[edit]
- ^ abcdWeilbacher, Michael (March 2012). 'Dedicated Servers in Gears of War 3: Scaling to Millions of Players'. GDC 2012. GDC Vault.
- ^Bernier, Yahn (2001). 'Latency Compensating Methods in Client/Server In-game Protocol Design and Optimization'. Valve Corporation. Retrieved 17 September 2011.
- ^ abArmstrong, Andrew (2009). 'The Mammoth Dedicated Server Guidebook'. Mammoth Media.[permanent dead link]
- ^ abcFiedler, Glenn (24 January 2010). 'What every programmer needs to know about game networking'.
- ^ ab'Source Multiplayer Networking'. Valve Developer Community. Retrieved 13 February 2013.
- ^'HTTP Fast Download'. Steam Support. Valve Corporation. Retrieved 5 May 2012.
- ^'Server Plugins'. Valve Developer Community. Valve Corporation. Retrieved 5 May 2012.
- ^Bowling, Robert (21 October 2009). 'Modern Warfare 2 For PC'. Archived from the original on 21 October 2009.
- ^Denton, Jon (July 4, 2012). 'Battlefield 3: How Fan-Run Servers Are Ruining DICE's Game'. Eurogamer.
- ^Walker, Robin (12 March 2009). 'And that's what ye git fer toochin that!'. Team Fortress 2 Blog. Valve Corporation.
Part of a series on |
Internet hosting service |
---|
Full-featured hosting |
Web hosting |
Application-specific web hosting |
|
By content format |
Other types |
A dedicated hosting service, dedicated server, or managed hosting service is a type of Internet hosting in which the client leases an entire server not shared with anyone else. This is more flexible than shared hosting, as organizations have full control over the server(s), including choice of operating system, hardware, etc. There is also another level of dedicated or managed hosting commonly referred to as complex managed hosting. Complex Managed Hosting applies to both physical dedicated servers, Hybrid server and virtual servers, with many companies choosing a hybrid (combination of physical and virtual) hosting solution. There are many similarities between standard and complex managed hosting but the key difference is the level of administrative and engineering support that the customer pays for – owing to both the increased size and complexity of the infrastructure deployment. The provider steps in to take over most of the management, including security, memory, storage and IT support. The service is primarily proactive in nature.[1] Server administration can usually be provided by the hosting company as an add-on service. In some cases a dedicated server can offer less overhead and a larger return on investment. Dedicated servers are hosted in data centers, often providing redundant power sources and HVAC systems. In contrast to colocation, the server hardware is owned by the provider and in some cases they will provide support for operating systems or applications.[citation needed]
Using a dedicated hosting service offers the benefits of high performance, security, email stability, and control. Due to the relatively high price of dedicated hosting, it is mostly used by websites that receive a large volume of traffic.[citation needed]
- 2Bandwidth and connectivity
Operating system support[edit]
Availability, price and employee familiarity often determines which operating systems are offered on dedicated servers. Variations of Linux and Unix (open sourceoperating systems) are often included at no charge to the customer. Commercial operating systems include Microsoft Windows Server, provided through a special program called Microsoft SPLA. Red Hat Enterprise is a commercial version of Linux offered to hosting providers on a monthly fee basis. The monthly fee provides OS updates through the Red Hat Network using an application called Yum. Other operating systems are available from the open source community at no charge. These include CentOS, Fedora Core, Debian, and many other Linux distributions or BSD systems FreeBSD, NetBSD, OpenBSD.
Support for any of these operating systems typically depends on the level of management offered with a particular dedicated server plan. Operating system support may include updates to the core system in order to acquire the latest security fixes, patches, and system-wide vulnerability resolutions. Updates to core operating systems include kernel upgrades, service packs, application updates, and security patches that keep the server secure and safe. Operating system updates and support relieves the burden of server management from the dedicated server owner.
Bandwidth and connectivity[edit]
Bandwidth refers to the data transfer rate or the amount of data that can be carried from one point to another in a given time period (usually a second) and is often represented in bits (of data) per second (bit/s).
95th percentile method[edit]
Line speed, billed on the 95th percentile, refers to the speed in which data flows from the server or device, measured every 5 minutes for the month, and dropping the top 5% of measurements that are highest, and basing the usage for the month on the next-highest measurement. This is similar to a median measurement, which can be thought of as a 50th percentile measurement (with 50% of measurements above, and 50% of measurements below), whereas this sets the cutoff at 95th percentile, with 5% of measurements above the value, and 95% of measurements below the value. This is also known as Burstable billing. Line speed is measured in bits per second (or kilobits per second, megabits per second or gigabits per second).
Unmetered method[edit]
The second bandwidth measurement is unmetered service where providers cap or control the “top line” speed for a server. Top line speed in unmetered bandwidth is the total Mbit/s allocated to the server and configured on the switch level. Unmetered bandwidth services usually incur an additional charge.
Total transfer method[edit]
Some providers will calculate the Total Transfer, which is the measurement of actual data leaving and arriving, measured in bytes. Although it is typically the sum of all traffic into and out of the server, some providers measure only outbound traffic (traffic from the server to the internet).
Bandwidth pooling[edit]
One of the reasons for choosing to outsource dedicated servers is the availability of high powered networks from multiple providers. As dedicated server providers utilize massive amounts of bandwidth, they are able to secure lower volume based pricing to include a multi-provider blend of bandwidth. To achieve the same type of network without a multi-provider blend of bandwidth, a large investment in core routers, long term contracts, and expensive monthly bills would need to be in place. The expenses needed to develop a network without a multi-provider blend of bandwidth does not make sense economically for hosting providers.
Many dedicated server providers include a service level agreement based on network up-time. Some dedicated server hosting providers offer a 100% up-time guarantee on their network. By securing multiple vendors for connectivity and using redundant hardware, providers are able to guarantee higher up-times; usually between 99-100% up-time if they are a higher quality provider. One aspect of higher quality providers is they are most likely to be multi-homed across multiple quality up-link providers, which in turn, provides significant redundancy in the event one goes down in addition to potentially improved routes to destinations.
Bandwidth consumption over the last several years has shifted from a per megabit usage model to a per gigabyte usage model. Bandwidth was traditionally measured in line speed access that included the ability to purchase needed megabits at a given monthly cost. As the shared hosting model developed, the trend towards gigabyte or total bytes transferred, replaced the megabit line speed model so dedicated server providers started offering per gigabyte.
Management[edit]
Dedicated hosting services primarily differ from managed hosting services in that managed hosting services usually offer more support and other services. As such, managed hosting is targeted towards clients with less technical knowledge, whereas dedicated hosting services, or unmanaged hosting services, are suitable for web development and system administrator professionals.
To date, no industry standards have been set to clearly define the management role of dedicated server providers. What this means is that each provider will use industry standard terms, but each provider will define them differently. For some dedicated server providers, fully managed is defined as having a web based control panel while other providers define it as having dedicated system engineers readily available to handle all server and network related functions of the dedicated server provider.
Server management can include some or all of the following:
- Operating system updates
- Application updates
- SNMP hardware monitoring
- Application monitoring
- Application management
- Technical support
- Firewall services
- Antivirus updates
- DDoS protection and mitigation
- Backups and restoration
- Database administration
- Performance tuning
- Software installation and configuration
- User management
- Programming consultation
Dedicated hosting server providers define their level of management based on the services they provide. In comparison, fully managed could equal self managed from provider to provider.
Administrative maintenance of the operating system, often including upgrades, security patches, and sometimes even daemon updates are included. Differing levels of management may include adding users, domains, daemon configuration, or even custom programming.
Dedicated server hosting providers may provide the following types of server managed support:
- Fully managed – Includes monitoring, software updates, reboots, security patches and operating system upgrades. Customers are completely hands-off.
- Managed – Includes medium level of management, monitoring, updates, and a limited amount of support. Customers may perform specific tasks.
- Self-managed – Includes regular monitoring and some maintenance. Customers provide most operations and tasks on dedicated server.
- Unmanaged – Little to no involvement from service provider. Customers provide all maintenance, upgrades, patches, and security.
Security[edit]
Dedicated hosting server providers utilize extreme security measures to ensure the safety of data stored on their network of servers. Providers will often deploy various software programs for scanning systems and networks for obtrusive invaders, spammers, hackers, and other harmful problems such as Trojans, worms, and crashers (Sending multiple connections). Linux and Windows use different software for security protection.
Software[edit]
Providers often bill for dedicated servers on a fixed monthly price to include specific software packages. Over the years, software vendors realized the significant market opportunity to bundle their software with dedicated servers. They have since started introducing pricing models that allow dedicated hosting providers the ability to purchase and resell software based on reduced monthly fees.
Microsoft offers software licenses through a program called the Service Provider License Agreement. The SPLA model provides use of Microsoft products through a monthly user or processor based fee. SPLA software includes the Windows Operating System, Microsoft SQL Server, Microsoft Exchange Server, Microsoft SharePoint and shoutcast hosting, and many other server based products.
Other software applications available are specialized web hosting specific programs called control panels. Control panel software is an all inclusive set of software applications, server applications, and automation tools that can be installed on a dedicated server. Control panels include integration into web servers, database applications, programming languages, application deployment, server administration tasks, and include the ability to automate tasks via a web based front end.
Most dedicated servers are packaged with a control panel. Control panels are often confused with management tools, but these control panels are actually web based automation tools created to help automate the process of web site creation and server management. Control panels should not be confused with a full server management solution by a dedicated hosting providers.
Limitations[edit]
Many providers do not allow IRC (bots, clients or daemons). This is due to rogue IRC users triggering DDoS attacks against the provider, which may overwhelm their networks, lowering service quality for all customers.
- Adult content is disallowed by many providers as it may either be of questionable legality or consume large amounts of bandwidth.
- Copyright violations – Hosting copyrighted material of which an individual does not own the copyright to is against the terms of service of most hosting companies.
See also[edit]
References[edit]
- ^“Internet Infrastructure Technology” Structure Research, Sept 5, 2012