|
|
High Availability
Swift utilise the highest quality Oracle Sun Servers to ensure exceptional reliability, however if the worst happens and hardware fails, our High Availability Failover system will ensure that downtime is kept to an absolute minimum.
The servers are monitored 24/7 and engineers automatically paged within 15 minutes of a fault being detected.
If your server fails due to a hardware failure, we can simply mount your container on the SAN from the spare hardware, and boot the container up. This failover process only takes a few minutes with automated scripts, and the data is completely up to date. In comparison, restoring from a back-up takes many hours, and all data since the last back-up is lost.
 |
 |
UK Network. The networks high availability UK backbone comprises of a heavily redundant 10 Gigabit network across nine major London data centres with automated failover to alternate paths in the event of a network segment failure. See seperate diagram.
|
| Local Data Centre Network. Servers are connected to the backbone via two local data networks with automated failover in the event of a firewall, switch, or cable failure; to ensure high availability at the local network level. |
Servers. Each hardware node (physical server) hosts a number of virtual servers. Each virtual server is connected to the Storage Area Network (SAN) and does not use local hard drives.
In the event of a physical server failure, the server may be mounted to the SAN from an alternate hardware node and started normally. The advantage is that the downtime is minimised to minutes, there is no loss of data and data is completely up to date.
The diagram illustrates the failure of servers 6-10 on hardware node 2, which are simply mounted and started on hardware node 4. |
Storage Area Network (SAN). The storage area network is the key to our hosting platforms high availability. It has been designed with no single point of failure, and can withstand multiple simultaneous failures.
Two dedicated networks are employed for the Storage Area Network, which automatically fail-over in the event of a network failure. Each server is connected to both SAN networks. The SAN Storage server is fitted with two disk controllers which automatically fail-over in the event of a disk controller failure, with each network connected to both controllers. This eliminates any single point of failure, and can even withstand the failure of a network AND controller simultaneously with no loss in service or performance. The disk array employs RAID, and can withstand multiple simultaneous disk failures. |
| Back-up Storage Server. Each virtual server is backed-up every night to a storage server. The backup may be used to restore a whole server to a hardware node, or individual files. At least 10 days worth of backups are achived. |
|
|
| |
|
|
