Does Real-Time Backup and Recovery Exist Today?
As long as cyber attacks and viruses continue to infiltrate computer networks, and as long as systems continue to fail for a myriad of other reasons as well, anxiety over data protection and recovery will remain at or near the top of the list of major concerns for IT managers across the globe.
The costs associated with data loss are staggering. According to the National Archives and Records Administration, Washington, DC, 93 percent of companies that lost their data center for 10 days or more due to a disaster filed for bankruptcy within one year of the disaster. And 50 percent of those businesses that found themselves without data management for this same time period filed for bankruptcy immediately.
In addition, according to the 7th Annual ICSA Lab’s Virus Prevalence Survey (conducted in 2002), recovery from desktop-oriented disasters costs the average company between $100,000 and $1,000,000 per year. These figures include both hardware and software costs.
Even with all of the backup and recovery solutions available in today’s market, the question that keeps popping up is whether or not real-time back-up and recovery exists today.
The Debate over Real-Time Backup and Recovery
Phil Goodwin, senior program director for the META Group, seems to think that it does not. He believes that real-time backup and recovery will not arrive until storage virtualization matures, which he foresees as not occurring until 2005 or 2006.
Jon Toor, director of marketing at ONStor, believes that Mr. Goodwin is correct in that some form of virtualization is required for real-time backup and recovery. After all, he says, the intent is to provide uninterrupted access to data, irrespective of where the data is located. “But virtualization is a broad term, and it implies a level of complexity that is not essential – and perhaps not even desirable – for a backup and recovery role,” asserts Toor.
An example, he says, would be the ONStor filer. In this environment, servers and clients on the LAN store filed-based data to a SAN filer, which then stores it to SAN-attached disk. “On a scheduled basis, the SAN filer replicates the data to a second disk array. With a second data copy to draw on, data accesses may be transitioned from the primary to the secondary array at any time. The client may be unaware of this transition. In this sense, it is a form of virtualization, but a simplified form that is technically accessible now,” says Toor.
Zophar Sante, vice president of marketing for SANRAD, believes that real-time replication does exist today through virtualization. “Real-time replication is done through advanced mirroring techniques from the network layer, meaning that intelligent SAN gateways have the ability to replicate incoming writes and replicate them in a synchronous real-time fashion to independent storage systems located within a SAN,” he says.
Sante adds that if a write packet comes into a SANRAD virtualization switch, it is converted to a FC write request and replicates the write request, sending it to individual storage systems within a FC SAN architecture. “We are doing this today even if the storage systems are in different buildings or even across town connected using FC direct or FC tunneling. This technique means that data is replicated in real time to a remote storage system, and because these remote systems are generally disk based, their performance is equal to that of the primary system,” he concludes.