By Heidi Biggar
It has been several years in the making, but end users are beginning to embrace the concept of continuous data protection (CDP), a new category of data protection that industry experts believe has the potential to change the face of “backup and recovery” as we know it today.
According to Mich Fisher, co-chair of the Storage Networking Industry Association’s (SNIA) recently formed Continuous Data Protection Special Interest Group (CDP SIG), “What CDP is, from a standards point of view, is one of the more intriguing debates going on right now.” Formed in February, SNIA’s CDP SIG is working to define CDP, as well as developing CDP standards. One thing members do agree on is that “CDP is all about the recovery point objective [RPO]-that is, being able to have any recovery point along the line,” says Fisher.
What differentiates CDP from existing backup-and-recovery techniques-including traditional tape and disk backup, mirroring, snapshots, and replication-is its ability to provide near-instantaneous restore to virtually any point in time (right up to the point of failure) without having to physically move or copy data (see figures below).
Also, unlike traditional snapshot technologies, which require users to predict how often failures will occur and then set “appropriate” protection levels (e.g., snapshot frequency) to minimize overall risk, CDP assumes failures can occur at any moment and therefore continuously protects data to ensure recovery at any point up until the actual failure, explains Jeff Wells, vice president of research operations and co-founder of Diogenes Analytical Laboratories, an IT consulting and product evaluation firm in Erie, CO.
As for mirroring, Wells says issues with data consistency can “blow RPOs” with this type of technology if users have to refer to a snapshot to ensure the data that is being restored actually matches the original data.
Benefits of CDP include near-instantaneous recovery of data, lower costs (fewer processes and copies of data to manage), better use of disk resources (e.g., ability to allocate high-cost disk resources to online applications), and simplified management (see “Objectives of data protection,” p. 10).
Founding members of SNIA’s CDP SIG, which is part of SNIA’s Data Management Forum (DMF), include Alacritus, EMC, Hitachi Data Systems, InMage, Mendocino, Mimosa Systems, Network Appliance, TimeSpring, Revivio, Scentric, Storactive, Sun, and XOsoft. The DMF has established a hierarchy of data-protection technologies (see figure below).
Following are two examples of how early adopters are using CDP technologies from Revivio and Storactive to continuously protect data in server and desktop/laptop environments, respectively.
Baptist Memorial Health CareImplementation: Revivio CPS 1200
Hal Weiss, a systems engineer at Baptist Memorial Health Care, in Memphis, TN, says he stumbled onto the concept of CDP when he was doing some research on data protection last year. “We have a huge amount of data that needs to be protected, and we were looking for something that would solve our backup problems,” Weiss explains. “Our ability to recover data affects people’s health.”
Baptist Memorial’s storage environment currently includes more than 360 million financial and patient information files on 30 Unix and Windows servers spread across 15 locations. According to Weiss, traditional tape backup methods were causing problems for the healthcare provider. For example, in the case of one 9TB server, backup took as many as 30 days to complete.
After reviewing various replication products and a couple of CDP options, Baptist Memorial decided on Revivio’s Continuous Protection System (CPS) 1200 along with Veritas’ Volume Manager software. (Volume Manager is necessary for the Revivio implementation.) Behind the CPS appliances are HP EVA disk arrays.
Weiss says that he particularly liked the fact that the CPS 1200 was both storage- and OS-agnostic and that it did everything at the block (versus file) level.
Baptist Memorial implemented two Revivio CPS 1200 systems in February and moved production applications onto the devices last month. It plans to move an Oracle database as well as its payroll system to the CPS 1200 in the near future.
Although implementation of the product has been seamless, Weiss says that transitioning from a traditional tape-based backup infrastructure to the CDP system has been slow due to some hesitancy within his organization to change architectures.
“We’re still doing traditional backup,” says Weiss. “Tape is hard to break away from, despite its problems.”
Rather than using a series of frozen images, which can consume large amounts of media space (tape or disk) to restore data, Revivio relies on “time-stamped” records to minimize storage capacity requirements and to provide near-instantaneous access to data. Weiss expects to see a return on his $312,000 investment ($300,000 for the two CPS appliances and $12,000 for Veritas Volume Manager) within 12 months.
On the downside, each CPS only supports two applications (or timelines); support for four and eight applications is reportedly in the works.
At last month’s Storage Networking World show, Revivio demonstrated how its CPS 1200 can be used for both single-item recovery and with replication for class-A disaster recovery (data was replicated between two CPS units). Revivio introduced its first CDP product (known then as TimeFrame Data Protection System) in October 2003.
Vitale, Caturano & CompanyImplementation: Storactive LiveBackup
A “disaster” situation prompted Vitale, Caturano & Company, a mid-sized CPA firm in Boston, to investigate data-protection options for its network of 300 laptop/mobile computers. Until then, the firm had no data-protection policy or software in place for its mobile users; it used Legato’s NetWorker on the server side.
“One of the firm’s partners lost all his data when his machine crashed,” says Jonathan Holmes, technology manager at Vitale, Caturano. “And he was one of those users who saved everything locally, not to the network.”
Holmes says that while the firm did look into traditional backup options, as well as a local backup provider and an application services provider (ASP), the firm ultimately decided to implement Storactive’s LiveBackup for Windows PCs because it allowed the company to keep data on-site, made recovery easier and near-instantaneous, and cut down on storage media costs by making/storing only one copy of each file on the server, according to Holmes.
Changes can be transmitted over a remote connection to the network or stored in local cache if the network connection is unavailable. LiveBackup uses a series of “continuous recovery checkpoints” to ensure that data is as recent as the last save, or any previous saved version.
Since implementing Storactive’s LiveBackup, Holmes says he has only had to do a full system recovery twice, but he says that individual file-recovery capabilities are tested on average twice a week. Currently about 75% of the CPA firm’s 300 laptops are running LiveBackup. Holmes says that he would like to have all the firm’s laptops running the software, but because of the initial “seeding” process, which can take up to 24 hours to complete, he has had to roll out the application one laptop at a time.
Objectives of data protection
Protection against different types of threats
- Storage or infrastructure failure
- Logical attack or data corruption
- Accidental over-write or deletion
Minimize exposure to data loss
- Recovery point objective (RPO)
Shorten time to recover operational data access
- Recovery time objective (RTO)
Reduce application downtime during data protection
- Data-protection window