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

IBM Global Services has reviewed DCH's requirements document and has prepared a solution based on DCH's assessment of application criticality. IBM can match techniques for data replication, storage, retrieval and restoration to meet the stated recovery levels needed to protect DCH's business and ensure availability of its critical applications.

For companies conducting electronic business, IBM offers e-business Recovery Services, a comprehensive portfolio of services to protect organizations from disruptions to critical business functions.

To satisfy DCH's most critical applications--electronic-commerce infrastructure--IBM recommends IBM High Availability Services. This consists of a dedicated configuration of Sun Microsystems servers for the Web, application and database servers, and a Compaq Computer Corp. server for the IVR (interactive voice response) system. Internet access will consist of full-time Internet connectivity using VPN (virtual private network) technology at 128 Kbps for transmission of replication data that will grow to 24 Mbps of Internet connectivity during a disaster event.

Dedicated systems will serve as the repository for real-time replication of the data and as the recovery systems for application and transaction processing should the DCH production systems be rendered unusable. IBM will provide a T1 to MCI's 800 service so that DCH can reroute inbound calls to the recovery IVR through MCI's command routing feature, and a remote-access router with 20 dial lines to reconnect with the payment processors and trading partners. Server equipment equivalent to the production environment will ensure the recovery center is able to step in and process user requests to the same degree as the production system within the recovery-time objectives specified.

For DCH's second major recovery requirement--the Intranet Web server and NT file servers--IBM will make Compaq systems available from the recovery center resource pool with the necessary software loaded within four hours of a declaration using the IBM NT Recovery Management Services. Access to these applications for DCH operations personnel can be onsite at the recovery center or through remote connection at our Chicago local access site. Remote users will have access to these servers within 24 hours through an MCI frame relay network connection.

Internet access at the recovery center is through high-speed frame relay access to the IBM Internet backbone. This connection is implemented over a 45-Mbps physical access on diversely routed Sonet-protected optical fiber.

To ensure that DCH's business-recovery plan remains current with its production environment, we have included 48 hours of test time in the first year and 24 hours in the remaining years.

Network Computing's Evaluation of IBM's Proposal

First things first: IBM made an early commitment to our RFP process and it was the only respondent to deliver its response on time, without requesting an extension.

In IBM's proposal, DCH would configure database replication to propagate changes that would essentially mirror its current servers within the IBM hot site. That new configuration would create four copies of the data, one each in the active and standby servers at DCH's site, as well as copies in the active and standby servers at the recovery hot site.

IBM economized on startup and ongoing costs by leveraging existing Internet connections. IBM says it would work with DCH and its ISP to ensure QoS (Quality of Service) by guaranteeing only one ISP between DCH and IBM. However, IBM points out that since transaction volumes are driven in large part by Internet clients, any slowdown or outage on the VPN would be offset by a corresponding dip in Web page hits. In other words, if the sessions supporting replication are affected, the likelihood of a significant queue of changes building up is remote because the same outage will affect client connections as well.

Granted, use of VPNs for mission-critical paths is not as questionable as it was once regarded. In this case, both organizations already have a high-speed connection to the Internet, and a VPN connection would add little cost. However, any channel across the Internet is a shared facility, so DCH would be wise to scrutinize propagation and latency delays in the VPN channel as well as increases in the total bandwidth requirements.

IBM's solution essentially requires DCH to duplicate its investment in Internet servers. Left as is, without further explanation or any enhancement, one can only imagine the reaction on the part of management: "We need to double our investment in servers, maintenance and associated manpower, and all we get is disaster recovery?" On the plus side, however, IBM's solution carries no declaration fees--in other words, DCH can use the equipment at the hot site whenever it wants, for as long as necessary, without extra expense.

Brian Walsh is the founder of bwalsh.com, a Portland, Ore., consulting firm specializing in Internet and client/server product strategies, development and testing. Send your comments on this article to him at brian@bwalsh.com.