And ost plug-in installed the nbu media server

Symantec NetBackup (NBU) Design
Best Practices with Data Domain

Introduction

Data Domain provides an alternative nearline storage solution for NBU customers who are faced with never-ending data growth and unabated storage expansion associated with ballooning amounts of backup and archive data. While NBU is one of the most scalable data protection solutions available to the market, data growth and data retention requirements drive near-continual expansion of NBU storage resources.

The scope of this whitepaper focuses on how the Data Domain deduplication storage solution integrates with standard NBU architectural and operational environments in order to overcome the growing gap between what is actually being accomplished and what needs to be accomplished.

Figure 1: Common NetBackup Architecture

3

Typical NBU Challenges

The typical NBU environment supports anywhere from a handful to thousands of clients. NBU scales by

• Eliminating redundant data backups (copies of databases, aggressive backup retention policies,

etc.)

Technology Overview

Data Domain reduces unnecessary NBU data storage via inline data deduplication and traditional

5

significant differentiator from competitive products which perform deduplication at the file level or at a block level.

6

Data Domain Architecture and Models

Table 3: 2007 Data Domain Systems, Addressable and Logical Capacity

Note: Logical Data Storage Values above reflect deduplication and compression effects on backup data. The actual values are dependent on rate of change and backup policies.

File System and VTL Integration

Data Domain systems support two integration methods with NBU, either via network file system mounts or as a standalone Virtual Tape Library (VTL). Data Domain systems can run in a mixed mode capacity, providing both interface methods concurrently to one or many NBU media servers. This flexibility affords a great number of integration scenarios for NBU. The following figure illustrates both integration scenarios with NBU media servers.

The VTL interface emulates a STK L180 tape library, and requires a fiber-channel connection along with the appropriate NBU device driver. NDMP backups are supported only via the Data Domain VTL interface. Multiple instances of VTL can be created per Data Domain system instance. Up to 64 LTO tape drives, 10,000 slots, and 100,000 virtual cartridges can be created per Data Domain system instance. As a standalone VTL, existing physical tape resources can be leveraged by native NBU capabilities.

Replication

Figure 5: Collection and Directory Replication Modes

Note: Some NBU environments are architected to support extremely high-performance backups for high-volume clients. Typically, specialized designs are implemented to support backup speeds of 1-4 TB/hour. The Data Domain system architecture can be configured to support high-performance workloads (via multiple parallel instances) with each Data Domain system instance supporting 200 MB/sec aggregate workloads per controller on currently shipping Data Domain systems. Because Data Domain’s product architecture is CPU-centric, this number typically changes (upward) with new product releases in a given price band. The top end of the Data Domain controller line, with dual-socket Intel CPU components, has gone from 40 MB/sec. (DD200 in year 2004) to more than 200 MB/sec (DD580/g in year 2007), a factor of five increase over three years. Please check current Data Domain literature for current platform names and throughput.

Planning / Sizing Considerations

Deduplication benefits are realized over time and eventually plateau once the backup versioning policy and the incremental backup traffic is fully realized. Since the characteristics of data vary by data type and production environment, a combination of backup policies, data change rate, and data structure impacts Data Domain system sizing estimates.

Sizing

Note: For customers using VTL mode, Data Domain systems do not require space pre-allocation for virtual tape volumes. As virtual tape volumes are mounted and filled in the Data Domain system, physical space is not ‘hedged’ for scratch virtual tape volumes. Other VTL technologies pre-allocate physical tape volumes, regardless of whether or not they are empty, full, or active.

An initial sizing metric for the Data Domain system is to tally up the primary volume of data on all backup clients, or a 1:1 ratio of primary data to the Data Domain system’s addressable capacity. Then run backups against the Data Domain system. Depending on retention policies, compression rates, change rates, etc, three to six months of backups can be retained on disk. As an example, an environment with 5TB of data requiring backups would need a Data Domain system with 5TB of storage. In this case, three to six months of backups could be stored on a single 5TB Data Domain system.

The following figure illustrates NBU integration with a Data Domain system in a primary and alternate site configuration. In this architecture a single NBU master / media server is shown. In a production environment, a single NBU media server can be successfully mapped to multiple Data Domain system instances, or vice versa with the appropriate NBU licensing. Note that the customer becomes ‘DR Ready’ much faster with this inline deduplication and replication approach compared to systems that do deduplication as a post-process or where tapes are physically transported off-site..

Figure 7: Mapping Data Types to Backup Directories and Data Domain System Instances

14

If a Data Domain system instance reaches capacity, individual directories can be migrated to a new Data Domain system instance using replication, with minimal reconfiguration and downtime required. For example, if all Oracle, Exchange, and file server backups are mapped to individual Data Domain system directories, a specific backup data type can be migrated to a new Data Domain system to reduce capacity in the original Data Domain system instance, allowing for additional capacity and growth of the remaining data types.

Large NBU environments contend with a never-ending challenge of performance optimization for system resources. To avoid performance bottlenecking and to obtain optimal Data Domain system performance, multiple Data Domain systems can be deployed to support parallel workload / performance demands. At present, each Data Domain system presents a separate Storage Unit for NBU.

A significant aspect of all data protection solutions is the ability to send important data to a distant location where the intent is that it will be kept safe from any disaster that may affect the primary data site. That data can then be used to restore operations within the tolerance of a specified Recovery Point Objective (RPO) and Recovery Time Objective (RTO). Offsite storage solutions can also play a large part in data archival, and NBU allows for several duplication methods for both measures.

If a tape-only environment is configured to push enough data to keep multiple drives spinning, inline copy is an option for duplication. This option can be configured to make up to four (4) online copies of the backup as it is being taken. The speed of these copies, however, is limited by the slowest resource each copy is being directed toward.

• The Vault option allows for a completely automated system and frees administrator time to concentrate toward other tasks

The OST plug-in is also another method of controlling the deletion and replication of backup images from a single point. OST allows NBU, versions 6.5 or higher, to better communicate with disk-based storage devices. This option is available to NBU customers that utilize a Data Domain system, allowing NBU to integrate with and take advantage of the benefits provided by the Data Domain systems.

Utilizing the advantages of the disk based backup destination that a Data Domain system presents allows a NBU administrator to efficiently set a timeline for daily operations, and meet that timeline. These daily operations consist of the typical routines, outlined by the following diagram.

17

Figure 10: NBU Daily Operations with Data Domain and Tape Duplication

19

Note: The Data Domain system cleaning process can be resource intensive. A ‘throttle command’ is available to assign the relative priority to cleaning processes compared to normal backup and restore I/O. We recommend the Data Domain system cleaning processes run periodically as part of standard NBU operations, scheduled when backup/restore activity is at minimum levels. See the DD OS Admin Guide for more discussion.

20