Every aspect of the Weberz network was engineered with two primary goals in mind, stability and redundancy. This is due to the network infrastructure being the most important component of the Weberz Hosting operation. Everything from the backbone Internet connections to the routing protocols used were chosen specifically to achieve these two goals.
Our data center is connected to the Internet using two OC-48 fiber connections that were dedicated to the Weberz Hosting network. Each of these fiber Internet connections are connected to separate self healing fiber rings that are owned and operated by separate transport providers. This was specifically designed so the Weberz network was not reliant on any one service or transport provider. Over these fiber connections we can provision any number of separate Internet service providers to distribute and balance our network load. Once in the data center the fiber lines are converted into usable Internet connections and terminated into the Weberz Routing systems.
The routing system design for the Weberz Hosting network was specifically engineered for maximum redundancy and minimum down time. All of the routers used in the system design are powered using OpenBSD. The Weberz routing system utilizes OpenBGPD software to provide its multi stage platform using eBGP and iBGP services.
- Each Internet service provider is terminated into a separate provider level router. This router’s sole purpose it to provide connectivity to its designated provider and perform eBGP with the provider. The eBGP service downloads a full list of routes for the Internet to be sent to the Service Routers.
- Every Service Router has a iBGP connection to each of the available Provider Routers. Each Service Router downloads the full Internet routing table available from the Provider Routers. These routes are then sorted and the best provider path is selected for each route on the Internet.
- The Service Routers also announce to the Provider Routers the Weberz IP address blocks with a scaling metric number that sets the Service Routers priority in the Service Router fail over sequence. These IP blocks are then re-announced to the Internet Service Providers, making the Weberz IPs available on the Internet.
- Each of the Service Routes also have network wide firewalls and static routing paths to control the access and flow of data into the Weberz network. They are then connected to separate network switches to allow for maximum redundancy. Each of those switches connect directly to every other switch on the Weberz Network including each other.
This routing system allows reduced down time in the event of a hardware failure by allowing each of the routers to be actively connected at all times. When a Service Router fails the network IP blocks are automatically re-routed to the next Service router in line within seconds. Once the failed router comes back online, it re-establishes connectivity to all the systems, collects all the current routing and packet state information, and re-assumes its place in system.
As mentioned in the Routing Systems section above, each of our primary service routers are connected to a separate networking switch. These switches are known as the Master Switches. Each of the master switches are then connected to each rack switch in the Weberz Network and to all of the other master switches. Using a protocol called spanning-tree, within seconds these switches automatically fail over to the next master switch available in the event of a hardware failure.
All of the network switching hardware on the network is made by Dell or Cisco systems.
Shared Hosting Cluster
Unlike most hosting companies that put all of an account’s resources on a single server, Weberz puts each service for an account on a separate physical server. This means your databases are stored and processed on a separate system than your website. Weberz also separates your email on to a different physical server that does nothing but process and filter email. This helps the overall performance of all of the web services on your account and balances the load on to separate servers.
By separating these services on to different servers our customers can be sure that a website going offline is not going to cause their email to go offline or vice versa. Weberz also uses separate physical servers for DNS, Time Servers, Black Lists, and our control panel system. An example of our shared hosting cluster would look like this: