SRDF, WDM Spell Disaster Recovery for EMC Corporation

Two trends recently drove the demand for solutions, such as EMC's Symmetrix Remode Data Facility (SRDF). First, the old method of disaster recovery become unfeasible. Second, enterprises became more and more dependent on data; no longer could they let revenue be jeopardized by the threat of a disaster eliminating a data center.

EMC Corporation did not have to be educated on the value of a sophisticated disaster-recovery strategy. That’s a class EMC is well experienced in teaching.

A provider of intelligent enterprise storage systems, software and services, EMC sought to implement its own product, Symmetrix Enterprise Storage, to deliver the same level of protection for mission-critical data it had provided to other companies.

The problem: limited optical-fiber capacity between two data centers.

The solution: a Wavelength Division Multiplexing (WDM) system from ADVA Optical Networking.

The result: EMC’s product development and other mission-critical data is continuously, completely mirrored via its Symmetrix Remote Data Facility (SRDF) application. At the same time, Asynchronous Transfer Mode (ATM), Gigabit Ethernet and other high-speed, local area network application traffic passes along the same fiber backbone. All protocols run at their native speeds.

"The Race" Is Off

Before SRDF emerged in 1994, information-centric enterprises had one basic strategy for recovery from data center disasters. IS departments dubbed it "The Race to Sunrise," and here is how it typically went:

• Databases were copied to tape, a four- to six-hour process ensuing at midnight, so as not to interfere with the opening of the following business day.

• Tapes were stacked for delivery to a secure site after the start of business for the organization’s shipping department or a contract transport service.

• Tapes were driven to the remote site and secured (two to four hours).

The best starting point for recovery from a data center disaster was already 12 hours after the last backup. Then came the lengthy, labor-intensive recovery procedure:

• Appropriate recovery tapes were identified at secure site (two to four hours).

• Tapes were transported to the recovery site (two to four hours).

• Data was restored to disk (eight to 18 hours).

• Databases were rolled forward to the best recovery level (six to 12 hours).

If things went as smoothly as could be hoped, application processing resumed a whopping 18 hours after the data-center disaster. Data was at least 30 hours old.

Two trends drove the demand for solutions. First, the old method of disaster recovery became unfeasible as commerce became more global in nature and the business day expanded across time zones; windows for data backups effectively evaporated. Second, enterprises became more and more dependent on data; no longer could they let revenue – now reaching millions of dollars per hour – be jeopardized by the threat of a disaster eliminating a data center.

The dated practices became obsolete. Sophisticated strategies emerged.

The SRDF Effect

In the five years since the solution was launched, more than 5,000 SRDF licenses have been installed globally. It is the unquestioned industry leader.

In a typical implementation, Symmetrix Enterprise Storage systems are deployed at two sites. SRDF, a host-independent solution, mirrors production or "source site" data on a separate recovery or "target site" in realtime. The process is transparent to users, applications, databases and host processors.

The SRDF solution positions enterprises to more quickly recover from earthquakes, floods, fires, hurricanes, protracted power outages or destruction of primary processing facilities by terrorist activity or hundreds of other unforeseen causes. When the primary system is down, high availability software signals a quick switchover to the copied data at the target site. An enterprise’s critical information is available in minutes.

Business operations resume full functionality with minimal interruption. After the source site is restored, via repair or replacement, SRDF resynchronizes data to ensure information and database consistency.

The solution also enables enterprises to accommodate a variety of more conventional, planned network events such as daily backups, database loads and refreshes, application testing, scheduled maintenance and data center migrations and consolidations. Enterprises have come to regard the network outages associated with each of these as unavoidable, mundane, cost-of-business issues; in fact, they are major impediments to business continuity, and they are eliminated by SRDF. Backup operations and new application testing are simply performed from the target site, while normal operations are conducted from the source. Throughout, information is protected and available.

In these ways, SRDF delivers three categories of benefits to enterprises:

Business. Customer service is enhanced because of reduced network downtime. Recovery from data center disaster is faster, increasing the availability of revenue-generating applications. Longer runs of transactional applications boost revenue. Additional business opportunities are created through the greater availability of data and more frequent intervals of practices such as data warehouse loads and refreshes.

Operational. The streamlined disaster-recovery procedure eliminates the slow, labor-intensive tape-loading and retrieval schemes. More frequent and less costly online testing of disaster recovery scenarios provides IS staffs with greater confidence in a viable and flexible solution. Simultaneous support for heterogeneous operating systems simplifies IS-staff training requirements. Built-in synchronization functionality preserves database integrity.

Financial. Expensive, inflexible manual backup and restoration procedures are eliminated. Resource utilization is maximized because a single SRDF target site can support multiple source sites, each with multiple hosts.

The benefits are so distinct that the return on an investment in SRDF is often realized days after implementation.

EMC’s Situation

EMC decided it was time to take advantage of its own technology. In 1998, EMC decided to implement its own solution to ensure that it could efficiently recover from a calamity at its primary data center in Hopkinton, Mass. – that the company’s 11,200 employees worldwide would maintain immediate, easy access to customer, product and corporate data 24 hours a day, seven days a week.

What didn’t appeal to EMC was the prospect of a possible 18-month turnaround to get permits to run additional optical fibers between its source and target sites. The backbone of 24 fibers between the two sites was exhausted with ATM, Fast Ethernet and Gigabit Ethernet application traffic. EMC required a solution that would make use of only the existing fiber infrastructure and yet support SRDF and Enterprise System Connectivity (ESCON) traffic, in addition to the in-service applications. It also wanted to position its network to support additional, faster protocols as the company’s networking needs evolved. Furthermore, it sought a protocol-transparent solution – one that did not require EMC to expend resources to physically replace modules or configure data rate changes.

WDM – requiring only an active wavelength division multiplexing system at each facility – was the clear choice.

The technology had emerged in the early 1990s to enable improved bandwidth performance over dark fiber. Metropolitan financial institutions were the first enterprises to deploy private optical-fiber based networks between facilities as a high-bandwidth solution for recovering from data center disasters, storing data, clustering high-speed computers for parallel processing, handling the facility moves associated with mergers and acquisitions, implementing multimedia business tools, etc. WDM is ideally suited for such applications, and many SRDF customers have already implemented WDM solutions.

WDM multiplies the capacity of optical fiber strands, creating "virtual channels." Incoming application traffic is converted to specific wavelengths and multiplexed. Multiple channels are transmitted over the same fibers. There is no performance degradation or specific protocol requirements because light waves of different lengths do not interfere with one another during transmission, and are converted back to their original formats at output. Consequently, across distances up to 50 kilometers, enterprises can safely, simultaneously utilize ATM, Coupling Link, ESCON, Ethernet, Fast Ethernet, Fiber Distributed Data Interface (FDDI), Fibre Channel, FICON, Gigabit Ethernet, SONET, SRDF, Sysplex Timer and other high-speed, proprietary applications.

Choosing Wisely

EMC evaluated a variety of vendor solutions and selected ADVA’s Optical Channel Multiplexer (OCM) 8 and OCM 16 systems – sold as the OM/9000 Model 25-8 and Model 25-16 by INRANGE Technologies. These were two proven, highly-experienced solutions providers. ADVA has grown its installed base to more than 150 service-provider and private-enterprise customers in its first five years, and INRANGE, with personnel in 50 countries, is an industry leader in sales, service and technical support of data center-networking, channel-extension products.

The ADVA solutions EMC purchased are eight- and 16-channel WDM systems that lend the company the ability to expand capacity simply and pay for the additional bandwidth as needs evolve – one of the primary criteria in EMC’s decision-making process. Price was a significant factor, as well.

Footprint and survivability also weighed in EMC’s decision. The ADVA OCM 8 and OCM 16 systems both feature 19-inch, rack-mount, hot-swappable, modular architectures. Other systems EMC considered required a significantly greater amount of the valuable real estate in the company’s source and target data centers. Some had stiffer power requirements or less flexible environmental tolerances.

Ease of implementation and use were considered, and the ADVA solution – featuring plug-and-play installation capabilities and requiring little maintenance – graded well. Implementing the ADVA solution placed minimal burden on EMC’s network support staff. Since then, EMC has managed and monitored the WDM system via its existing software management platform – made possible by the ADVA solution’s SNMP software-management module.

Conclusion

Running the SRDF application across the WDM links, Symmetrix Enterprise Storage systems continuously mirror EMC’s mission-critical data at each Massachusetts site. Such an event, thankfully, hasn’t occurred, but, if a data center disaster knocked out either the source or target site, EMC would experience very little interruption before business operations returned to normal. EMC network support staff has proven as much, by running multiple, successful in-service tests of what is not so much a "disaster recovery" as it is a "disaster restart" strategy.

At the same time, EMC has multiplied available bandwidth by up to 16 times – giving itself room to grow, without investing in expensive, time-consuming upgrades of its optical-fiber infrastructure. ATM, Fast Ethernet and Gigabit Ethernet application traffic traverses the same fibers with the SRDF signals, and all protocols run at their native speeds.

The final analysis, ADVA’s cost-effective, proven, flexible WDM solution has satisfied each of EMC’s needs – for today and tomorrow.