 | |||
|
|
||
| |||
| |||
| |||
|
||||||||||||||||||||
| |
|
Home > Data Backup News > The state of data backup in 2009, Part 4: Managing the backup and recovery process |
| |
|
|
|
|
| Data Backup News: |
|
||
|
|||
|
|||
|
|||
|
|||
|
|||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||
Reporting and classification tools can help, but face challenges Because of its size, financial resources, and business model, Yahoo Inc. has extensive in-house programming and application coding resources. These resources have allowed Yahoo's manager of data protection Marcellus Tabor and his team to develop a centralized custom reporting tool for both primary storage and backup. The reporting tool is based on a MySQL database with a PHP interface and draws in data through SNMP and other means. Yahoo had no choice but to develop it in-house because Tabor says he's yet to find any commercial offering that would be as effective.
"Backup vendors are going to have to get more intelligent about the way they do incrementals," he said. "If you have [a certain] kind of data set, incrementals can take longer than fulls." Backup software will eventually have to locate changed blocks of data without crawling an entire file system, Tabor predicted. "And any kind of parallelization they can add can't come soon enough," he added. Tom Becchetti, a storage administrator for a large medical equipment manufacturing company, said he's found Aptare Inc.'s StorageConsole helpful in getting a good sense of his backup environment. Aptare helped him find a network bottleneck that saved his company from adding tape drives. "It helped me quickly see that we're not pushing the tape drives and start asking why," he said. But Becchetti says he's yet to find a classification tool that's universally interoperable. "The thing about mainframes is you can apply security and management policy to the name of a file, it'll enforce adherence to the name across systems," he said. "The hurdle in open systems is that there are so many different operating systems and file systems. To get market share is very difficult." Archiving could cut the data glut Beth Israel Deaconess Medical Center storage architect Michael Passe said traditional backup paradigms will lose their usefulness when storage grows to the petabyte level. "We're going to have to shift gears," he said. "The systems you're starting to see come out for that amount of data are focused on low cost per gigabyte, internal redundancy with clustering, and geographic replication." Preparing for that shift, Beth Israel has purchased an F5 Networks ARX Series file virtualization switch to automate the movement of unstructured data between tiers of storage for backup and archiving purposes. Automated migration will help make that task more manageable with large data repositories, Passe added. "Going forward, the focus will be on keeping active data on primary storage, while anything that doesn't change should come out of the data stream," he said. The plan is to use the Acopia switch to segment out 180 days of data to be retained on a Data Domain box as an archive. Passe said he's aware of the risk of shifting vendor lock-in from a tape drive maker to the file virtualization switch vendor, but said the tradeoff is greater flexibility in his choice of backup and archiving products. Ultimately, improving process may be the only way out Becchetti said his background in mainframe gives him a different perspective on the data growth currently happening in the open-systems world. "In the mainframe world, backups are backups, and you backup for two weeks and only for two weeks," he said. "The shift in open system is that you backup incrementally every day, a full weekly, and once a month you send a backup off site forever." Changing this will require more of a change in mindset than technology, Becchetti said. "I've done things to try to bring up end users awareness, like cutting down on backups of home share," he said. "Next year, I hope to do away with backing up home shares, and if someone needs a file backed up, they'll have to put it in the department share." Patrick Banghart, manager of the Windows server team for a large health insurance company in California, was told to revamp his company's backup environment when he was hired in 2007. At the time, his company was using an outdated version of CA Inc.'s ARCserve software for tape backup. Banghart is in the process of deploying backup to disk using Symantec Corp. NetBackup 6.5's DSSU feature. While the majority of his company's data will be backed up to that repository before being sent to tape, Banghart says database application data will still be sent straight to tape. "Database data tends to stream well, and it's faster not to have to move that type of data twice," he said. Banghart's classification system consists of dividing unstructured and structured data based on the type of application server being backed up. "Data classification is a challenge," he said. "But it's critical to moving forward in any kind of efficient manner." The next step is to identify data owners and begin the mentality shift. "Previously, capacity has just been provided, without a predetermined need by the IT staff to know who owned it or how long it should be retained for," he said. "IT people in operations, rooted in the practical and the tactical, haven't asked." They're asking now, he says, because "data growth is making it critical -- growth is basically expense. In our case, we're reaching logical limits. Our NetApp filer is maxed out and we have to rethink the environment before it becomes even more painful." Pacific Gas and Electric Company (PG&E) data center storage team lead David Ping says it's difficult to get application owners and his storage team on the same page. "The storage group knows a lot about storage, and the application groups know their environment," Ping said. "I'm not seeing a person that understands both or looks at the entire environment regularly." Ping has put together a one-to-two-hour class bringing in host systems administrators to tell his storage team about things they do that impact storage subsystems, and so the storage team can talk about how subsystems impact the host. "As always, though, this is just one of the many projects in front of us," he says.
|
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
About Us | Contact Us | For Advertisers | For Business Partners | Site Index | RSS |
|
|
||
|
|||||||
