This section explains the individual components required for the hosting solution, as set out in the flowchart displayed in "For the technical."

Network Switches

The network switches facilitate the linking of multiple devices on the PRIAM Co-Lo network.

Redundancy is taken care of by two redundant switches so, in the case of failure, one can be swapped out without any system downtime.

Cisco PIX 515’s

The PIX515 takes care of all VPN connections between PRIAM Co-Lo and remote locations.

The PIX’s are configured in failover mode which means if one fails the other will take over automatically with minimal loss of service: end users should not notice this.

The configuration should be backed up every time it is changed and stored in a central location.

Firewall

The firewall is the single most important component of the entire PRIAM hosting setup; without it nothing can function. It is responsible for monitoring and securing all connections between the internal networks and the outside world

The firewall has two hard drives configured in RAID1 for resilience against a hard drive failure. There are also redundant power supplies to provide resilience against a single power supply failure. The machine also has a DRAC card which provides console access to the server from a remote location and even allows power cycling. This server also benefits from DELL Support.

The firewall ensures traffic security by filtering and traffic shaping. This is done by limiting and prioritising all traffic that passes through it. The firewall also monitors and stores detailed statistics on network traffic and provides graphical representations of these statistics.

Every night the firewall configuration files are backed up and copied to the weekly backup location at PRIAM-HQ. The firewall is also responsible for taking care of the backups for the other servers at PRIAM Co-Lo.

Redundancy for the firewall is provided by means of a 2nd firewall server that is configured identically to the main firewall and is tested on a regular basis. In the event of a failure the faulty firewall is swapped out. All cables are tagged and colour coded so this can be done by remote hands onsite for minimum down time.

 

Database Servers

Each database will have a primary and a shadow server. In the event of failure of the primary server the shadow server will become the primary server and the failed server will be repaired or replaced and take up the role of shadow server.

All servers are identical and have the hard drives configured in RAID10 for maximum drive performance and resilience. There are also redundant power supplies to provide resilience against a single power supply failure. The database servers also contain a DRAC card, the same as in the firewall. The database servers benefit from DELL Gold Support for maximum reassurance.

Every night the database is backed up and then zipped up to the backup folder along with the server configuration files. These files are then copied by the firewall and placed in the relevant backup location.

 

Web Site Load Balancer

The website load balancer handles all web requests and forwards them on to one of the web servers, determined by the load balancing policy. This method makes the web servers very easily scalable: if we need more performance we can simply add another web server with no down time.

The load balancer is not directly connected to an external interface. For security reasons all traffic is routed via the firewall.

Redundancy for the firewall is provided by means of a 2nd server that is configured identically to the main load balancer and is tested on a regular basis. In the event of a failure the faulty load balancer is swapped out. All cables are tagged and colour coded so this can be done by remote hands onsite for minimum down time.

Web Servers

There are multiple web servers with all websites on all servers. A web server receives a request via the web site load balancer.

Redundancy is taken care of by the load balancer and the number of web servers that are in operation. As long as there is at least one web server running there will be no interruption to service. The load balancer will automatically detect temporary failures (network issues, system crashes, server overload, and technical intervention) and remove the related servers from the pool. Then, at regular intervals, it will probe the servers that have been marked as down. As soon as they come back to life they will get added to the pool again. It doesn't matter if 1, 2 or 10 servers are down, clients will still be able to access the sites.

All servers will benefit from RAID1, redundant power supplies and DELL Support.

Every night the web site directory gets zipped up along with the server configuration files; this backup is then copied by the firewall to the relevant backup location.