TCO analysis: where D2D fits—part 1

Computer Technology Review, July, 2004 by Tony Asaro

There are thousands of storage systems that are being used for disk-to-disk (D2D) backup and it appears that nearly every customer is considering implementing a solution within the next 18 months. Backing up to disk is one of those things that just makes sense. The price of hard drives is decreasing while capacities increase; data growth is off the charts; backup windows are shrinking; reliable recovery of data is requisite; and D2D technology is advancing.

The Enterprise Strategy Group recently completed a report, The Evolution of Enterprise Data Protection, which surveyed over 220 IT managers. Based on this research, the majority of these users are using or will be deploying D2D solutions of some variation. The benefits are clear:

* Increased backup performance

* Decreased time to recover

* Ensured data integrity and recoverability

* Media management issues

Some of the reasons that customers were interested in deploying D2D solutions:

* "One problem we have is that we can't recycle backup-tapes fast enough. This has really affected our ability to backup and restore desktops. Our backup application backs up all daily changes across thousands of desktops so tapes get filled up pretty quickly. It sometimes takes us three days to restore a desktop because the user's data is scattered across 40 different backup tapes!"

* "With tape-based systems, the human overhead of managing tapes is much more pronounced than with disk-based products."

* "We really only want to use tape for disaster recovery purposes. We'll address data retention with cheaper, more readily accessible disk technologies."

* "A big problem we have is that we have no way to know whether or not a backup tape is good until we try to recover."

The question is not whether to implement a D2D solution but what options are available, and which will solve the problems to meet your needs.

Backup to Disk (B2D)

On the surface, backup to disk is very straightforward. Most major backup software applications allow users to backup to a disk target. Customers did not back up to disk in the past due to the high cost of disk storage. Today low-cost, high-capacity ATA drives makes backing up to disk economically practical. But backing up to disk is not as simple as it would seem. There are issues that IT managers should be aware of.

The file systems on the media server qualified by the backup vendors typically have a 2-TB limitation. This will impact scalability, management and performance. You can get around this by installing a third-party file system but this will add cost and complexity. Using a host-based file system may expose you to viruses, file corruption, and accidental or malicious file deletion.

The storage array used for the backup must be configured and managed and comes with all of the same technical challenges as any other SAN-attached system. Imagine managing 50TB of backup data. Based on the 2TB file system described, this would mean that the storage administrator would have to support 25 file systems. And if you configure a LUN per file system, that equates to the management of 25 LUNs. Backup polices will more than likely need to be reconfigured to backup to disk.

Having said this, there are a number of users that are doing exactly what is described above. In the ESG report, a large number of customers are using ATA drives in their storage arrays and are performing backup to disk. Right now, this solution is good enough for their purposes. But as the demand and storage capacity grows, IT managers should look at special built D2D solutions, including backup to disk (B2D) appliances and Virtual Tape Libraries.

NetApp and Data Domain are two vendors that provide B2D products that resolve some of the issues described above. NetApp supports up to 16TB in a file system; Data Domain supports 1.5TB (most customers want to keep data sets small to maintain decent recovery performance) and provide powerful compression capabilities that allow customers to effectively store more data with less actual used storage capacity. ESG Lab tested Data Domain and found that they can easily achieve 20:1 compression.

Both NetApp and Data Domain have developed products that have their own file system. This eliminates the issue of file corruptions, viruses and file deletions. Data Domain claims that their file system was designed to recover from major disasters allowing users to rebuild data with just the drives.

Both products are NAS systems, which are usually easier to install and manage than a SAN storage array. Backup polices will still need to be modified to work with these solutions; however, customers that have done this say that the changes are minimal.

Virtual Tape

Virtual Tape is not a new concept and has been widely used in IBM Mainframe environments for years. ESG research has found that approximately 60% of all IBM Mainframe data is protected by virtual tape. Virtual Tape solutions now support other operating systems including Windows, Unix and Linux.

Virtual Tape Library (VTL) solutions are storage subsystems that emulate a tape library and use RAID-protected hard drives to store the data. Typically the VTL solution has its own file system, emulates multiple tape libraries, and provides additional performance boosts through load balancing and large block sizes.