High Availability / High Reliability / Our Infrastructure

Elephant Outlook’s inherent performance and reliability allows Elephant Outlook to offer service level agreements at levels far in excess of standard environments.

A hardware role is composed of a group of features. It is a unit that logically groups the required features and components needed to perform a specific function in our environment.

Elephant Outlook’s hardware infrastructure includes some of the following roles:

• Mailbox Information Stores.  These are the back-end servers that host mailbox data for users.
• Client Access. This is the middle-tier server that hosts the client protocols, such as Post Office Protocol 3 (POP3), Internet Message Access Protocol 4 (IMAP4), Secure Hypertext Transfer Protocol (HTTPS), Outlook Anywhere, Availability service, and Auto Discover service. The Client Access Servers provide the connectivity and access to our services for users.
• Hub Transport. This is the mail routing servers that routes mail.
• Edge Transport. This is the mail routing servers that typically sit at the perimeter of the topology and routes mail in to and out of the Elephant Outlook network.
• Hygiene. These servers handle anti-virus and anti-spam management.
• Mobile Messaging. These handle mobile access services like RIM Blackberry Enterprise and other mobile access servers for device synchronization.
• Systems Management and Monitoring. These servers handle the monitoring, score cards, server level agreement reporting, and management.

Failed disks, memory, processors, power, and network equipment are all common sources of unplanned downtime. Both Server Cluster and NLB provide availability in the event of a failure of a processor, memory chip, power supply, or other hardware component.

Clustering completely eliminates downtime associated with deploying patches by using rolling upgrades. Elephant Outlook performs a rolling upgrade by taking redundant cluster nodes gracefully offline, upgrading them, and then bringing them back online. This process is repeated until all cluster nodes have been successfully upgraded.

Cluster continuous replication (CCR) is a high availability feature that combines the asynchronous data shipping and replay technology with the failover and management features.

CCR is designed to provide high availability for Exchange 2007 Mailbox servers by providing a solution that addresses the following requirements:

• Mailbox solution with no single point of failure.
• Data availability and resiliency to failure.
• Service availability and resiliency to failure.
• Backup data center support with very current data.
• Rapid recovery with current data to any single failure.
• High availability for Mailbox servers.

Our SAS drive enclosures are dedicated and directly attached to each mailbox role and to each mailbox cluster node (total of 2 arrays for each cluster pair) to deliver speed, performance and reliability that satisfies demanding server applications as Exchange where data is actively stored and frequently changing.

Each enclosure utilizes multiple RAID controllers that perform the necessary calculations, controls I/O operations, and communications. Given their critical tasking, the enclosures are designed to incorporate twin active/active RAID controllers with mirrored cache for high-availability. If one controller should fail, the surviving controller can handle the full processing load until its mate is brought back online.

SAN is a network of shared storage devices, which included our high-availability EMC CX3 Ultra fiber disk arrays and Overland Robotic tape libraries built to prevent any single point of failure.

The architecture of our SAN is such that storage resources are made available to multiple servers and multi-paths on our network. Because stored data does not reside directly on any of the network's servers, processing power is optimized and storage capacity is provisioned on the server where it is needed most.

Using the latest equipment from Juniper Networks, Elephant Outlook utilizes multiple routers for redundancy connected to multiple tier-one providers as Level3, FPL Fiber, Progress Energy, and Time Warner Telecom. Each respective path to our providers utilizes diverse fiber routes, SONET rings, and duel building penetration to ensure the highest degree of reliability and security.

To provide dynamic routing and the fastest possible data transfers, Elephant Outlook runs the Border Gateway Protocol (BGP4) for best case routing with our own ARIN assigned IP Network Address Space. This allows for the routing of packets of information sent out from our network to be evaluated and sent over the best route possible.

Because of our redundant network architecture, packets may be sent via alternative routes even if they are being delivered to the same end user. Should one of our providers fail, packets leaving or entering our network are automatically redirected through another route via a different provider.

Next: Mobile Access