As far back as IT managers can remember, enterprise data backup has always been a time-consuming, expensive, and sometimes unreliable and frustrating process. And as organizations continue to generate more and more data, IT managers need more and more time to back it up.
So, what’s an IT manager to do? In Part I of this article, we asked a few industry experts what advice they could offer in terms of solutions for achieving the goal of backing up more data in less time. And advice they gave us. Here are even more suggestions for solving the age-old dilemma of how to back up more data in less time.
Deploy a Data Archiving System
Peter Hunter, product marketing engineer at Nashua, N.H.-based EqualLogic, suggests that IT managers deploy a data archiving system. These solutions integrate with applications such as e-mail servers and database servers to trim out older data and move that data from primary storage to secondary or even tertiary storage devices. “This relieves the primary backups by essentially removing data from the primary storage, thus reducing the backup window and recovery times,” says Hunter.https://o1.qnsr.com/log/p.gif?;n=203;c=204660765;s=10655;x=7936;f=201812281308090;u=j;z=TIMESTAMP;a=20400368;e=i
Network-Level Data Mirroring
Some industry experts contend that with network-level data mirroring, a backup window is no longer needed. Creating independent mirrors of data across the network provides higher I/O performance, serves as a disaster recovery solution, and offers an architecture where the production data is never offline during backup. “Network-level mirroring is performed by a storage solution in a similar fashion as a RAID controller,” says Zophar Sante, vice president of marketing at SANRAD, a San Francisco-based provider of IP storage network solutions.
The major difference, he explains, is that RAID controllers mirror storage devices with a single enclosure. But with today’s intelligent storage switches, mirroring can be done at the network layer, according to Sante, because the switches create and maintain mirrored patterns/volumes anywhere within the network — even across traditional physical limitations such as enclosures and distance. And, he explains, local synchronous mirroring can now be performed between two or more enclosures.
Sante cites the following as an example: An intelligent switch in “Building A” with Fibre Channel-attached storage can keep the data files on the Fibre Channel (FC) storage synchronized with the FC-attached storage in “Building B.”
Because both storage systems can service I/O requests from clients, the mirrors can be temporarily split so that one mirror partner can be mounted by the backup software while the other partner continues to service I/O clients. “Once the backup is complete,” he says, “the mirror can rejoin its partner and be synchronized for the next backup and provide assistance with I/O requests and online DR capability.” Sante adds that this technique will allow data to be available 24x7 and allow backups to occur without the need for a traditional backup window.
Gigabit Ethernet and Dedicated Backup LANs
Another suggestion for backing up more data in less time is building a dedicated 1000 Mbps (megabits per second) Gigabit Ethernet (GbE) LAN for increasing backup performance. “Gigabit Ethernet ports and NICs are quickly approaching the cost of 10/100 Mbps interfaces of two years ago,” says Sante. “A GbE port can support 100 MBps (Megabytes per second), NICs costs about $100, and 24-port switches are available for a few thousand dollars.”
However, he continues, one of the problems with backup is the network speed across the LAN. “Users want to back up 100 MBs but restrict themselves with 1 MBps or 10 MBps bottlenecks because of old network connections,” says Sante. He suggests that by deploying a LAN specifically for backup traffic, users can see up to a 10x increase in the backup window.
Hunter agrees that the backup network itself often creates a bottleneck. He suggests replacing aging 10/100 Mbps networks with Gigabit Ethernet. “Gigabit Ethernet is inexpensive and readily available, and offers a solution that is guaranteed to improve backup performance.
Consolidate Backups to Data Center with Data Replication
Another way to increase backup performance and increase efficiency for branch offices and departments, according to Hunter, is to outsource the service to the data center core. “This can be accomplished by remotely replicating data from the branch office to the data center,” he explains.
Asynchronous replication is an effective way of accomplishing this without requiring significant WAN bandwidth requirements. “By keeping the backups at the data center, businesses can realize the economies of scale with a consolidated backup operations department,” he concludes.
Use Shared, High Speed, Block-based Storage Devices
Another alternative for backing up more data in less time is to eliminate the weakest link the backup process. In most cases, according to Sante, if it’s not the network or the tape drive, it’s the production disk drives.
Sante points out that the LAN and the tape drives are usually the first two elements that users need to address when tackling the problem of shrinking the backup window. Those two elements are followed by the host disk drives where the data for backup is located. Sante says that in most cases, more and more data needs to be backed up, but the servers are not able to move the data fast enough.
The key to increasing backup performance and general I/O performance, Sante says, is to migrate to a SAN. “This may be a big leap for many companies, but the new IP-based SANs are much more affordable and easier to install.”
Sante suggests that instead of organizations spreading out their budgets to maintain individual storage on each server, they can consolidate that budget by purchasing a more effective, high speed solution that can be shared by all of the servers. “Backup performance will benefit greatly by the increased performance of the centralized storage solution, and organizations will reap the added benefits of snapshots and network data mirroring.”
As the volume of data being produced in organizations across the globe continues to increase, the conundrum of backing up more data in less time will continue to be on the minds of every IT manager. The good news is that more effective and more efficient solutions continue to pop up to help solve this seemingly never-ending problem.