High Availability Design

High availability is an approach used to design systems and their associated service so that a given level of uptime will be met over the desired systems.

While all users and administrators seek maximum uptime, or availability, from the systems they use and administer, certain systems are critical enough that additional effort and money must be spent to minimize downtime. Hospitals, communications, and infrastructure control systems are the most common examples, but high availability systems design is used in many different industries.

All high availability design approaches improve system uptime. Applied to a single server, high availability design seeks to minimize the number of failure points in the physical system.

Improved Redundancy

Redundant internal hardware systems and high quality components minimize the degree to which hardware failures can bring down the system. Simple and highly robust software architecture minimizes the number of places where a bug or conflict can create downtime.

Advanced systems make use of sophisticated operating system design that allows patching and upgrading all components on a live basis, that is, without rebooting the server or compromising availability.

Since a single server represents a single point of failure, high availability systems generally include multiple servers behind a load balancer or which have failover capability. Load balancing distributes a given workload across multiple servers, so that a failure of one server simply redirects traffic to the other still active servers.

Failover is the ability to switch to redundant, standby servers or networks in the event one fails. In a system or network with failover capability, one server will regularly send a "heartbeat" signal to the standby server to ensure everything is running. If the signal stops or otherwise indicates the primary server is no longer functioning correctly, the second server will take over.

Increased Uptime

High availability clusters are the next major step towards increased uptime. In a cluster, many redundant computers are all capable of providing the same service. A redundant private network between the computers allows all of them to monitor each other, detect failures, and automatically configure and restart a given application on a healthy server without needing intervention from a server administrator.

Almost all websites can be made suitable for adapting to a web cluster, thanks to our expertise and custom web hosting control panel.

In an web cluster situation each component must be easy to start, stop, and kill, and its status must be easy to check. The service might also use shared storage systems, like NAS or SANs.

In a conceptual sense, high availability can be broken down into passive redundancy and active redundancy. High availability systems usually use a combination of both.

Passive redundancy refers to systems that have sufficient excess capacity to allow things to continue even if a given sub system fails. A boat with two engines driving two propellers has passive redundancy, as does an electric grid with two generating stations. A failure of one engine or generating station will reduce the boat's speed and the grid's capacity, but the boat will still keep going and the lights will stay on. Servers that use redundant power supplies internally make use of passive redundancy.

Active redundancy allows the system to keep going without a reduction in performance. Actively redundant systems will re-configure themselves so that there is no degradation in performance if a failure occurs. RAID arrays with hot spares as well as most high availability clusters are good examples of active redundancy.

icing budget usually plays a major constraint for startups and growing business. Dedicated server hosting can be a bit expensive than shared servers but if you are looking for better performance and customization options then the investment is worth every penny. Go for dedicated server hosting.

EZ1
  • Quad Core
  • 2.40 Ghz (Xeon X3220)
  • 2 GB RAM
  • 160 GB Storage
  • 2 TB Traffic
$100/mo* buy now How Much Can It Handle?
EZ3
  • Quad Core
  • 2.80 Ghz (Xeon X3460)
  • 4 GB RAM
  • 160 GB Storage
  • 2 TB Traffic
$230/mo* buy now How Much Can It Handle?
EZ5
  • Dual Hex Core
  • 2.80 Ghz (Xeon X5660)
  • 8 GB RAM
  • 250 GB Storage
  • 2 TB Traffic
$600/mo* buy now How Much Can It Handle?
EH1 Server Cluster
  • 2 x Custom EZ3 Servers
  •  
  • 4 TB Traffic
$790/mo* buy now
EH2 Server Cluster
  • 1 x Custom EZ1 Servers
  • 2 x custom EZ3 Server
  • 1 x custom EZ4 Server
  • 8 TB Traffic
$1865/mo* buy now
View all Dedicated Servers View all Server Cluster
*Residents of the European Union may be subject to Value Added Tax unless a valid VAT ID is provided.