At Web In A Box, we set out to deliver reliable and responsive web services-- to which we think we've found the perfect recipe.
Load Balancing
When a request comes into our network, it is passed to our load balancing servers. These servers constantly monitor the availability and performance of our servers and will direct your request onto one of those servers. In the event one of the servers becomes unavailable or over-loaded, the load balancer will start diverting queries away from that server, or stop sending queries to the server entirely. The load balancing servers themselves are also redundant, so if one load balancer becomes unavailable, the other server will detect this and take control of the balancing.
Our Server Clusters
High performance hosting requires high performance servers-- so rather than just having one server doing your bidding, we've deployed a whole cluster of them. Our hosting clusters are built on a combination of Ubuntu GNU/Linux and Windows Server 2003 operating systems and is deployed on Dell Poweredge servers. The configuration of our servers allows us to be extremely flexible and offer things like the ability to run either PHP4 -or- PHP5 on your site*, which you're able to change on the fly!
Data Storage
In a traditional storage network, all the customer data is generally stored on one large disk array. These disk arrays generally have inbuilt fault tollerance, but in the event that the disk array does go offline, -all- the customers data is unavailable.
To be honest, we didn't much like that idea, so to ensure the highest level of data integrity and reliability, our storage back-end is configured with disk replication to ensure in the event of a failure, your data is always available and more importantly, always safe.
When one of the servers in our cluster writes data to our storage servers, the storage server that receives this data will write it to it's disk array and will also send the data to the other storage server, which then also writes the data to it's local array as well. Once this is completed the storage server then signals the clustered server a signal to let it know the data was written to disk.
This means that we could have a catastrophic failure on one of our disk arrays, which would be automatically detected and our network would 'flip' and start using the remaining storage server until the first one is repaired.
Network Infrastructure
Our communications equipment is completely, 100% constructed with Cisco routers and Cisco Catalyst Switches. Configured with HSRP, our routers constantly monitor each other and if one of the routers becomes unresponsive, the other router will take over control of the network so that connectivity isn't lost.
Network Monitoring
At Web In A Box, we really like to know what's going on with our equipment, so you can be sure that we monitor every facet of our network-- From the basic stuff like CPU, Memory and Disk usage, to Mail connections, DNS queries and con-current web requests. We keep an eye on everything, to make sure things run smoothly.
Network Topology Map
Below is a logical map of how the Web In A Box network is constructed, but if you have any questions at all, Contact Us and we'll let you know.