The Crosby Data Centre
Home to hundreds of servers, the Crosby Data Centre and
office complex is based in the North West of England. Advanced security, fire
suppression, climate control and power protection equipment provide the ideal
operating environment for servers and network equipment.
|
|
We also have smaller facilities in Manchester and space at neutral
collocation centres in Leeds and London. These locations are primarily staging posts
on our network through which we connect with other service providers, ISPs and
telecommunications operators.
|
Having researched the potential problems that any network must be protected against, Crosby went
about designing a structured, sophisticated and robust network architecture. A well–structured
network is more manageable and scalable; sophisticated use of technology balances performance with
cost–effectiveness, and redundant servers contribute to a higher level of fault tolerance.
The three fundamental elements of the Crosby Data Centre architecture are:
- Performance and Reliability
The network's primary design criterion emphasises fast and reliable user access to services.
- Duplication and Segregation
From power supplies to network connections, duplicate resources and segregated paths combine to
avoid any single points of failure.
- Distributed Resources & Centralised Control
Equipment is dispersed, but managed from a centralised hub.
Data Network
Crosby owns and operates its own data network infrastructure. We have our own IP number assignments
and are consequently not reliant on any single company for our Internet connectivity.
Our network utilises fibre optic connections from BT, Telewest, Kingston Communications, Sprint,
Fibrenet, Redstone Communications, Level 3 and COLT. Our Data Centre has fibre presented from a
minimum of two different telecommunications providers, ensuring that no single point of failure can
exist within our infrastructure.
Throughout our network, key equipment is duplicated: we use technologies including BGP and Cisco's
"Hot Standby Routing Protocol" to ensure the continuous availability of a service even in the
event of the failure of one piece of equipment.
BGP4 provides multiple redundant internet connections for increased speed and reliability.
Data Network Features
- Non–internet connected gigabit network provides secure transit for data and ensures
that the internet–facing network cannot become congested.
- At least two routers running the "Hot Standby Routing Protocol" service for any
server to guarantee availability.
- All data is backed up to our tape library on a daily basis
- Multiple firewalls protect servers from malicious traffic, viruses and trojans.
- Replicated database servers are not Internet accessible for maximum security
- Redundant layer 3–7 switches provide load balancing 'fail over functionality'
and allow us to remove any server from the cluster to maintain it, without affecting your
services.
- A secondary web and file server cluster that replicates data from the primary cluster,
providing a service should our primary servers fail.
- A primary server and file storage cluster data is held on mirrored storage arrays.
Servers
We predominantly use Compaq and Dell servers running FreeBSD or SCO–Unix, all of which
have dual Intel Pentium Xeon processors, dual 100Mbit Ethernet adapters and a minimum of 2GB of
memory. Storage is spread across multiple Ultra 160 SCSI hard disk drives using Adaptec RAID
controllers.
Every server on our network runs a RAID based disk redundancy solution. Should any of the primary
disk drives fail, a secondary disk drive can take over the workload with no loss of
service&ndhash;availability. Each server also has a replication partner that can take over its
workload in the event of a problem or during planned maintenance. In addition, data on our
servers is backed up to our Storagetek Timberwolf DLT tape library, with tapes being stored off
site.
Feeds
The Crosby Data Centre brings fibre optic cable from multiple carriers into the facility. Fibre
is delivered from different telecommunication providers and enters the building at different
locations in physically separate paths that never converge. Dual fibre entry provides redundancy
at the carrier level and also protection from instantaneous loss of connectivity when heavy
equipment cuts a fibre line during the construction or installation of cable or gas lines.
Capacity
Our network is connected to the Internet via redundant, full–duplex fibre optic connections
to multiple backbone providers. Connections are directed by redundant BGP4 routers and VLAN
distribution switches.
The minimum connectivity for our fully dedicated servers is a 10Mbps network card with its own
dedicated 2Mbps connection.
Monitoring
We constantly monitor all servers on our network with a range of methods, from simple pinging
and web fetches to disk space and load checks. In the event of an anomaly or error, our on–call
systems' administrator is alerted via pager or SMS. Our advanced SMNP monitoring system
constantly monitors all services, including incoming telephone lines, connections to the mobile
operators for text services and 3G connectivity for our content feeds.
Fire Detection
| The Data Centre is equipped with an advanced fire detection and suppression system,
including FM200 (or equivalent) gaseous extinguishing systems, which provide rapid
discharge and flame suppression in the event of a fire while minimising damage to
equipment and reducing danger to personnel. |
|
Security
We take both physical and network security very seriously. We keep up–to–date with
the latest security upgrades, and our firewalls protect vulnerable points and block out known
unsafe networks.
Access to our servers is tightly controlled by both electronic and biometric means, through a
system of airlocks and secured areas. Full entry to the Data Centre also requires a final visual
check before any person can gain access to the equipment. Once in the Data Centre, activity is
monitored and recorded around the clock.
Features :
- Very early smoke detection apparatus (VESDA).
- CCTV system covering all entrances/exits and main areas.
- 24–hour video recording.
- Full perimeter alarms.
- A PAC security card access system.
- Visual verification for all persons attempting to enter the data floor.
- Redundant air conditioning systems.
- Steel mesh reinforced walls & steel doors.
- Motion sensors and security cameras.
Engineers are based on site to ensure the fastest response should any physical intervention be
required at any time.
Bandwidth
We operate a redundant, scalable and bandwidth on–demand network that connects us directly
to the internet backbone in both locations and is 100% powered by equipment from Cisco, the
acknowledged leader in internet networking equipment.
Our primary connectivity is fed to us by Sprint and Level 3 and every server is connected
directly to Cisco and 3com switches which in turn link to high performance Cisco VXR routers
connected to our bandwidth providers.
We are further connected to multiple internet exchanges ensuring the fastest path for all traffic,
and a high level of resilience in the unlikely event of a network failure.
Bandwidth scales according to load, ensuring that we always have enough to feed our customers
requirements, and we typically keep 60% of our bandwidth in reserve to ensure we are able to cope
with unexpected surges and peaks.
Power and Environmental Control
Power is supplied to the Data Centre from diverse locations and is backed up with an APC Symmetra
UPS and multiple standby generators. The Data Centre is equipped with sophisticated air
conditioning apparatus set at 20 degrees C and 50% relative humidity, to ensure that systems
operate as efficiently as possible.
Features :
- 2 x 1700 KVA Generator back-up @ N+1.
- An APC Symmetra UPS system - redundant parallel formation @ N+1.
- Close control downflow air conditioning units to guarantee stable temperature and humidity @ N+1.
- 2 independent AC systems DX & chilled water @ N+1.
- 2.5 MVA of UPS conditioned power @ N+1.
- Diverse Power Supplies with diesel generator backup.
- Provision for emergency natural gas powered turbo generators.
The Environmental Impact Of Computers
Crosby works with its technology partners to find new ways to improve the environmental
performance of its network.
In redesigning equipment and infrastructure to future proof our business, Crosby takes into
account the environmental impact that network assets such as servers, air conditioning and
cabinets can have. Examples of this include:
- Thinner service leads.
- Noise minimisation devices for server rooms.
- Slimmer masts for our broadcasting antennas.
- Wind and solar energy designs to contribute to our overall power use.
- Reduced power usage during off peak times.
Crosby has procedures, guidelines and standards in place for the installation of all equipment
and endeavours to design wireless internet access transmission sites that blend in with their
surroundings.
For more details of our efforts to reduce the impact of our operations on the environment, see
our Crosby and the Environment page.
Voice Over IP Network
Our VoIP systems are built around Crosby's global network – thereby offering Crosby the
cost–effective option of terminating a call to the PSTN network.
In many ways a VoIP call is travelling over the most redundant network known to man: the
Internet. Crosby owns and operates an extensive global backbone network including 23
international and domestic gateway switches located throughout Europe, the Americas and the
Asia Pacific regions. This network is supported by ownership in over 35 transoceanic fibre
optic cable systems, providing the connectivity of Crosby's global network to over 250
countries worldwide. A call on the Crosby VoIP system avoids multiple hops due to our
centralised routing and peering relationships, and Crosby routes and monitors a VoIP call in
the same fashion as a regular voice call. Each call has multiple termination options that are
selected based on quality.
|
