We've built our house of brick, so when the wind blows, you'll stay standing.

We base all of our services on a foundation of monitoring. It's the core of our proactive model.  Like building a house, if the foundation is weak, the house will collapse.  To prevent that from happening, we've built a support and monitoring infrastructure that is rock-solid.

Datacenters

Our core systems reside inside of an Equinix facility in New York.  Equinix is a carrier-neutral facility who builds their datacenters with consideration for green initiatives in power and cooling.  They focus on being very good at one thing - delivering space and power to colocation customers.

Our offsite systems are colocated in Amazon EC2, from which we provide services to other Amazon Web Services customers.

Network

Our core systems connect to the Internet through redundant Juniper routers, each of which is connected to two network providers: Level3 and Cogent.  We run BGP and have our own IP space from ARIN.  Our core switches and distribution switches are made by Cisco and connected in a fully-meshed topology with multiple paths between any two systems.  Our entire switching matrix is Gig-E, which we provide at no additional charge to our customers. The entire network is protected by redundant Juniper firewalls, and all access to the network for non-public services is via VPN.

Servers

All of our servers are fault-tolerant to the role they provide.  Each system runs as part of a clustered pair, either on different physical hardware or as virtual machines on different physical VMWare ESX nodes.  Each machine runs a clustering engine such as heartbeat or OpenAIS and uses DRBD for its clustered storage.  This provides resource failover in the order of a few seconds.  Our storage is a clustered SAN providing NFS and iSCSI services to all of the nodes.  No system role needs to be taken offline for maintenance - we work on the standby node of the cluster, then fail over to it and work on the other node.  All services remain fully operational throughout all maintenance windows.

Virtual Machines

We believe in the benefits of virtualization to leverage unused or underused hardware.  Our VMWare vSphere cluster runs on multi-core HP and Dell servers, backed by our clustered SAN storage.  Using VMWare's HA and VMotion technology, we're able to move nodes between hosts without taking them offline.  This means that in addition to the fault tolerance provided by the clustered service roles themselves, we have additional fault tolerance for moving virtual machines to other nodes when we perform maintenance on the VMWare hosts themselves.

Client Connectivity

All of our interaction with client networks is via VPN.  We provide hardware VPNs between our support network and client networks, and we enable customers to connect to our support infrastructure or their own via our Juniper Secure Access SSL VPN systems.  Any data which traverses the Internet is encrypted, with the exception of monitoring services which simulate client traffic (HTTP, SMTP, etc).  We believe in security, and we've made an investment in it so that you know our interaction with your systems is protected.