

Heading for Disaster?
January 11, 1999
In Comdisco's Own Words: Solution Summary
For an Adobe Acrobat format version of Comdisco's Disaster-Recovery Solution, click here.
In today's business environment, traditional recovery programs seldom achieve the recovery-time objectives and recovery-point objectives companies realistically require. Meeting these objectives is vital to DCH's operations.
As the leading provider of business-continuity services, Comdisco was the first to develop and implement high-availability solutions. Over the past decade, Comdisco has used these advanced technologies and processes to enable companies to reduce the time to restore systems and eliminate lost transaction exposure. Comdisco also delivers crucial recovery solutions for organizations that rely on the Internet for e-commerce/revenue-generating and customer-support activities. Strategic to our services is CCSNET, Comdisco's Sonet-based network that interconnects Comdisco's facilities and is devoted exclusively to business continuity.
Based on this foundation, Comdisco proposes a solution that ensures availability and integrity of each of DCH's environments.
For the most critical applications, Comdisco recommends a two-part strategy. First, DCH's standby systems will be relocated to our Chicago metro recovery facility to ensure that in a localized disaster DCH has access to its standby systems. Second, file systems and data will be replicated and transmitted to the high-availability services area of Comdisco's New York metro recovery facility. Here, a Sun Microsystems E450 manages the data in real time, and an EMC Symmetrix Array provides mirroring of critical information so DCH can conduct non-disruptive tests. Multiple preconfigured boot drives, ready to accept updates of operating system or application changes, will meet the low data-recovery-time requirements.
Additionally, a dedicated disk drive will be provided for DCH's IVR (interactive voice response) system, and dial-backup solutions will be provided for communications recovery with payment providers and trading partners. Internet traffic is redirected via UUNet and MCI using BGP4.
For DCH's midcritical systems, Comdisco recommends a hybrid hot-site approach. This combines daily electronic transmission of incremental data changes to Comdisco's New York facility. To ensure four-hour recovery, a dedicated disk will be preconfigured from full backup, thereby requiring that only incremental changes be applied. For DCH's high-tolerance network connections, we recommend MCI's frame relay backup PVCs.
Comdisco has applied proven services, plus creative use of technologies, to provide a cost-effective, technologically robust solution. This accommodates growth and migration to emerging technologies while meeting or exceeding the design criteria.
Network Computing's Evaluation of Comdisco's Proposal
Comdisco delivered the most extensive business resumption plan. What its proposed architecture lacks in elegance, as compared to IBM's VPN (virtual private network) solution or Exodus' load-balancing, it makes up for in attention to detail and in sheer capacity. This capacity translates to a channel for replication over a DS-3 circuit provisioned from DCH to Comdisco's Chicago Metro OC-48 Sonet ring. DCH's current standby servers are redeployed in Comdisco's Chicago facility. From there, data is transmitted over multiple links to Comdisco's recovery center in Carlstadt, N.J.
Comdisco uses simple components (such as preconfigured boot drives) to facilitate acceptance of replication on a scaled-back computing platform. This also eases the transition to actual recovery and validation, while still enabling performance within the low data-recovery-time arena.
Comdisco proposes configuring the servers with EMC Corp. Symmetrix Disk with business continuance volumes (BCVs) to provide the ability to validate/test in a nondisruptive mode. Comdisco provides a backup to the disaster-recovery servers. Before DCH could lose service, the company's own production site, and Comdisco's Chicago and New Jersey sites would all have to fail simultaneously. Each site is able to function independently.
|