The Boomerang Infrastructure

Boomerang has built an extremely secure and resilient infrastructure to ensure maximum availability at all times.

Redundant hardware has been installed at every tier so that the application can still continue to function even with multiple server or network device failures.

The network perimeter consists of a HA pair of Cisco PIX 515 Firewalls in an active - passive configuration.

In the unlikely event of a hardware failure at the Firewall level, a state full failover occurs; external connectivity is maintained at all times.

The firewalls are connected to the network via a pair of Cisco 3750 Switches in an active - active configuration.

In the event of a hardware failure at this level, no failover will be required as both switches are used at all times.

Servers are configured to use bonding / teaming to present both network interfaces to either switch, hence from the server level, no network outage is caused.

Before incoming traffic reaches the Web Servers it passes through a pair of Foundry Load Balancers in an active - passive configuration. The Foundry's will distribute the incoming connections in a round robin fashion to the two Web Servers.

This means that the load is distributed evenly between the Web Servers giving enhanced throughput and high availability.

Web Servers

Please see below server specification;

• HP DL360G5
• 2X Quad Core Zeon
• 16GB ECC RAM
• 1TB Local Storage using 2.5 SAS Drives in RAID 5 Configuration

A hardware or software failure at this tier would be detected by the Load Balancers, at which point all web traffic would be directed to the other functioning server.

Application/Database servers

We are using a further pair of HP DL360s using the above specification for the application and database tier.

Database replication will be in place to provide a local cluster along with external replication to our secondary data centre, 30 miles to the south of the primary data centre. Please see below for full datacentre overview.

Both Application and Database tiers provide full fault tolerance through application / hardware load balancing.

Data Centre

The key to the reliability of internet services depends almost entirely on the datacentre.

Our attention to detail in this area is the key to our reliability. In short, we provide the ultimate technical environment for the high availability of Boomerang.

Boomerang is hosted from our two self-managed Manchester datacentres; Reynolds House is our primary facility, and Greenheys is used as a backup/disaster recovery (DR) facility.

Reynolds House Datacentre

Reynolds House is located on the Manchester Technopark on the periphery of Manchester City Centre. Reynolds is a dedicated datacentre building; and in fact the largest high-specification datacentre building in the North West.

This is the primary location for the Boomerang infrastructure.

Greenheys Facility

Greenheys is our secondary facility for disaster recovery. The site is completely independent of Reynolds House ensuring that Boomerang is covered by an offsite backup facility.

We maintain a high level replication implementation for our secondary Microsoft database.

Two APC UPS systems enable us to maintain power to servers and networking equipment should the power fail. Fixed air conditioning in the server room allows us to keep the facility to an operational temperature of 22 degrees or below.

Superfast Network

It goes without saying that our network design is key to being able to provide you with 24/7 service.

That's why we've spent so much time optimising it for maximum speed and availability. Our attention to detail delivers a quality of service that is second to none.

Network Connectivity

We operate a fibre-optic metropolitan area network spanning four facilities in Manchester, connecting our two datacentre facilities to the internet.

We connect to multiple upstream bandwidth providers at three different Manchester datacentres, giving us complete independence from the failure of any bandwidth provider or facility.

The below diagram shows our network connectivity between facilities and the upstream providers to which we connect.

A note about resilience

When you're aiming for 100% uptime, resilience in the network is key. We have a very simple rule: if a device is critical to network availability, there will always be a hot-standby available.

All critical hardware, such as routers and firewalls is duplicated, meaning that the failure of an individual piece of hardware will have little or minimal impact on the overall running of the network.

We run BGP4 on our network, which means we pick the best routes to any given destination. In the event that the best route is not available, the next best route is chosen.

Monitoring

We take our network uptime and stability very seriously. As such, we have several levels of monitoring in-place to ensure that any issues are discovered at the earliest possible point, and specifically before they become service-affecting.

Every single router, switch, firewall and managed server is monitored internally. Any outages are notified to us instantly by e-mail and SMS to our on-call engineer.

In addition, we have external monitoring, which checks our critical routers and firewalls from five different worldwide locations.

For further peace of mind, we monitor saturation on links, with alerts being generated if critical links are acting unusually. This allows us to very quickly see any potential issues which may be arising, for example attacks, or compromised machines.

For more information, a free trial, or to apply for an account contact us now.