Hosting Infrastructure

Servers

The servers consist of an IBM BladeCenter chassis containing a number of individual blade servers. This design allows for maximum fault tolerance, quick and easy swap-out and upgrade options and reduction of single points of failure to almost zero. The blades themselves are effectively a server motherboard containing CPU, RAM and control components. No storage exists on the blades and there are minimal moving parts to ensure maximum resilience. This allows for any blade to be swapped out for an equivalent unit with no need for concern over the associated operating system, software and data storage. The BladeCenter chassis contains resilient power supplies and management circuitry to house the individual blades and it provides the connectivity to the network and to the storage systems. Each chassis houses up to 14 blade servers.

Storage

Storage for all the blade servers is provided by a Storage Area Network (SAN) based on IBM TotalStorage SAN hardware. This storage solution allows for complete fault tolerance by ensuring that all disks in use are members of one or more RAID arrays – all data is stored in at least two physical drives and is striped across multiple disks for maximum performance. Partitions from the storage array can be dynamically assigned to one or more of the blade servers, allowing for database server clustering and load balanced Web and application hosting solutions without any need for data replication between servers. Any disk failures result in a simple hardware swap-out with zero downtime.

The SAN infrastructure also allows for regular snapshot based backups of all data so that recovery of lost items is almost instantaneous. The SAN solution also facilitates future plans for full replication of the entire hosting solution to a secondary data centre to provide total fault tolerance even in event of physical attack on one of the hosting premises.

Connections from the BladeCenter to the TotalStorage SAN are via redundant Fibre Channel links providing maximum performance and full resilience. The TotalStorage enclosure contains dual SAN switches and dual power supplies which are fed by two independent UPS units with full secondary UPS power and generator power provided by the Global Switch facility. Initial storage capacity of the solution is approximately one terabyte with simple options to expand to any reasonable level of storage requirements.

Connectivity

Perimeter security of the hosting solution is dealt with by two Check Point firewalls in a load balanced high availability cluster configuration for total resilience and maximum performance. Load balancing of Web servers is facilitated by fully redundant Nortel Layer 7 switch modules within the BladeCenter chassis.

Internet connectivity is sourced in the form of IP transit from four independent tier-1 backbone providers within Global Switch and dual Cisco routers with BGP4 routing technology and a dedicated Autonomous System are utilised for total connectivity resilience. Senior has acquired its own Provider Independent IP address space from RIPE so we are truly independent of any single upstream IP provider.

Backups

Backups of the entire hosting solution are managed using a dedicated blade server running IBM TSM backup software. This facilitates comprehensive image backups both to local disk storage independent of the SAN and to an LTO tape library capable of storing up to 8 Terabytes of data. Tapes in the library are rotated off-site to the Senior HQ fire safe on a regular basis for disaster recovery purposes.

Management & Monitoring

An advanced monitoring system at Senior HQ runs multiple checks on the entire hosting facility every minute with alerts sent out by email to our support staff immediately should any problem be detected. Escalations are automatically sent out by SMS text message should any fault remain unresolved for longer than 20 minutes. This facility allows us to ensure we meet our SLAs and that we can spot in advance any issues with hardware, software, networking or traffic loads on our systems. All management of the hosting facilities is done remotely from Senior HQ via secure VPN connections across multiple internet connections for resilience. Any hardware issues detected are rectified by the hardware manufacturer on-site under 4 hour response service level agreements, and as we have no single points of failure on our network we are able to deliver on our industry leading levels of service availability and performance.