For an Adobe Acrobat format version of Exodu's Disaster-Recovery Solution, click here.

The proposed architecture for meeting DCH's business resumption requirements leverages Exodus' leadership as a solutions provider for mission-critical Internet applications. As a pioneer in Internet disaster-recovery planning, Exodus has developed a robust set of network, physi-cal and personnel resources for meeting DCH's business-recovery-plan needs.

The proposal places a secondary hot site in Exodus' Santa Clara, Calif., Internet Data Center facility. Web traffic is intelligently distributed between DCH's Chicago facility and the Santa Clara location with Exodus' MultiPath Net service. MultiPath Net will automatically detect a failure at one location and route requests to the second facility. The advantages of this load-balanced solution include:

- No downtime. Because both sites are continually operating, traffic is routed away from the inoperable location in the event of a failure. And because the backup location does not have to be brought online, the failover period can be reduced to minutes rather than hours.

- Reduced risk. A continuously operating secondary site reduces the risk that a cold failover site will not perform as planned in the event of a disaster.

- Efficient use of resources. Equipment and network resources at both locations are in operation rather than sitting idle as a backup.

- Improved performance. Exodus' solution has the added benefit of geographically distributing Web traffic across the network. Not only does the proposed solution route traffic away from a failed location, it also automatically sends users to the optimal Web site. The result is faster Web response.

Exodus' backbone network is a key component for connectivity. As a Tier 1 Internet provider, Exodus operates a nationwide DS-3/OC-3 backbone with connections to all major long-distance carriers and multiple public and private peering relationships. The strength of the network ensures that DCH's customers will be able to reach the Web site in the event of a failure that affects DCH's facilities or that has an impact on a portion of the public Internet. Exodus' backbone also provides a platform for replicating and backing up content between the two sites.

Network Computing's Evaluation of Exodus' Proposal

Use of the hot-site equipment is exactly what made the Exodus proposal unique. It reads as much like a description of distributed-processing, load-sharing architecture as it does a disaster-recovery scenario.

Exodus complements DCH's current capabilities with a hardened facility and advanced load-balancing routing based on Exodus' MultiPath Enterprise Agent, part of its MultiPath Site Service.

Using the MultiPath service, Web traffic will be shared across the primary facility in Chicago and a second location in Santa Clara, Calif. (at Exodus' Internet Data Facility). MultiPath normally distributes the network load between the two sites, but in the event of a site failure, MultiPath will detect the problem automatically and not distribute Web traffic to the failed location. Since both sites are continuously operating, the failover will occur almost instantly, resulting in a disaster-recovery solution with nearly zero downtime.

However, this places a considerable design and implementation burden on DCH. Since changes can be produced by either system, all data stores must support bidirectional replication. This is not a trivial task, but is fairly commonplace in high-end Web sites and is a prerequisite for any load-sharing network.

Exodus did not propose a backup for the disaster scenario; essentially no replication takes place if the primary site has gone down because the secondary site will always be kept up-to-date.

And finally, unlike IBM and Comdisco, both of which provisioned PC servers out of their own stock, Exodus leaves DCH responsible for supplying all the PC server components.