It is a fact that all hardware suffers from failure at one time or another. To prevent loss of availability and data it is therefore imperative to incorporate as much redundancy as possible to eliminate single points of failure. The CloudSigma cloud runs on custom infrastructure designed specifically for the task.

Hot swappable dual power supply units

Failure of a power supply unit has no effect on that server's availability. Failure of power supply units in servers is one of the leading causes of availability loss.

Redundant networking cards

Failure of one of the networking cards will not affect availability of that server. Each server has four separate outgoing connections to our switches.

IPMI management cards

All servers have IPMI cards allowing our technical engineers to constantly monitor every server in every rack. Additionally, in the unlikely event of a hardware failure or software freeze we are alerted automatically by email and SMS.

Fast 2.5inch hard drives

All client virtual drives are stored on fast laptop style 2.5inch hard drives which have much higher read and write times than typical larger (cheaper) 3.5inch drives. This translates to better disk performance for virtual server instances.

Drives in RAID 6 array

All hard drives are in RAID 6 arrays with hot swappable drives. All drives are monitored and any failed drive can be replaced immediately without loss of data or availability. RAID 6 tolerates the failure of up to 2 drives simultaneously without any loss of data or availability.

RAID controller cards with battery backed RAM cache

All our RAID arrays are managed via specialised hardware RAID cards. Our RAID cards contain a RAM cache of disk activity and are backed by battery. Our RAID cards allow us to reconstruct RAID arrays in the event of power or disk failure very quickly minimising performance and availability loss.

Dual 6 core processors

Virtualisation performance is enhanced with additional CPU cores. We have 12 cores in every server which translates to better virtual server performance. In additional users can manually specify up to 10 cores to be used in their virtual server instances.

No SAN?

That’s right, like Google, Yahoo and others we don’t use a Storage Area Network (SAN) but instead opt for a distributed storage solution. Distributed storage provides us with much higher levels of redundancy and resilience than any SAN solution could. SANs represent large single points of failure and require huge networking traffic. When used in the cloud this translates to major availability losses if a SAN fails (and they do!) as well as reduced virtual server performance and increased network lag. SANs make more sense for single enterprises and more traditional web hosting but not for Infrastructure-as-a-Service platforms such as CloudSigma’s.

Dual independent power supply

All components in the rack have dual power supplies and are fed from two truly independent supplies. In the event of loss of power from one supply (due to component failure for example), the rack will suffer no loss of availability.

UPS backed power supplies

Each of the two power supplies to each rack is backed by two independent Uninterrupted Power Supply (UPS) solutions. In the event of one or both power supplies failing, the UPSs would prevent any loss of power to the rack.

Redundant under floor cooling

Each rack is cooled by a redundant under floor cold air system maintaining the average temperature at 22 degrees Centigrade (72 degrees Fahrenheit). Keeping our racks cold increases reliability and performance.

Redundant switching

All switches within our cloud are redundant. Failure of a switch will not affect the availability of any components within our cloud.

Redundant routers

As well as redundancy in switching, we have redundant routers. In the unlikely event that a router suffers a hardware or software failure, our cloud will suffer no loss of availability.

Redundant IP connectivity

Our cloud is connected to the general internet via multiple IP connectivity providers. We have chosen providers that offer independent complimentary networks. In the event of a failure or scheduled maintenance at an upstream provider, we suffer no loss of network availability.

Fibre-optic networking

Our cloud uses fiber optic networking for high data flow connections in order to minimise network latency and maximise performance.