What you will learn in this tip: Whether justified or simply because the technology was misunderstood, backing up network-attached storage (NAS) devices
was once considered a challenge for backup and storage administrators. This tip reviews some of the most popular NAS backup strategies.
To put NAS backups into perspective, we must first look at a little history. Back when all disk storage was directly attached to the computer using it, backups were relatively straightforward; we installed data backup software that transmitted data to a directly attached tape device. As servers multiplied and environments grew larger, it became more manageable to transmit data over the network to a server that managed all backups and the required storage media. In fact, this remained a very common backup method regardless whether the disk storage was attached to the system via a SCSI cable or Fibre Channel. However, the growing popularity of NAS appliances required a fresh look at things.
Why NAS backups are different
NAS appliances may look like scalable file servers but they work very differently. For one, these appliances run proprietary operating systems that are optimized for storage performance. For that reason, you cannot install the usual backup software agents on a NAS device. Unlike Microsoft Windows, Unix, Linux, Novell or Mac OS-based servers which are widely supported by many backup software vendors, NAS devices require a different approach; let's review the main NAS backup strategies.
Traditional network-based backup
Traditional network-based backup relies on backup agents installed on all servers that access storage on a NAS device, and data is sent to a backup server across the network. This is essentially business as usual in terms of client-server backups, which is why it is referred to as "traditional network based backup" in this context.
While this approach requires no special changes to the environment, it is definitely not the most effective. For one, network traffic is increased as data must travel on the network from the NAS devices to the client and again from the client to the backup server. In addition, in cases where a NAS device is used as a file server shared by multiple systems, the backup software on each system can create multiple backup copies of the same files.
This situation can be somewhat improved by designating a single system as a backup proxy for the entire NAS device. In this context, using a proxy means that data backup for the entire NAS device is handled by a single system acting as a backup client. However, this method will not be effective in instances where NAS devices are also used to store application-specific data such as databases that require application-awareness or shutdown to generate useable backups.
Network Data Management Protocol (NDMP) is a protocol that was created specifically for NAS device backups. Put simply, it allows a NAS device to send data directly to a tape device or backup server across the network without the need for backup client intervention. In other words, the backup server communicates directly with the NAS appliance indicating which storage device (e.g., tape drive) data can be sent to for backup. Most mainstream backup software packages such those from BakBone Software Inc., CommVault and EMC Corp., as well as IBM Corp. Tivoli Storage Manager, Symantec Corp.'s Backup Exec and NetBackup now support NDMP and offer various levels of integration and functionality.
However, NDMP backups are best suited for file data -- and other than Oracle Secure Backup, which is NDMP capable -- are not well integrated with applications such as MS Exchange, SQL, DB2, etc., which need to be aware that a backup is taking place to ensure consistency. While this can be achieved through custom scripting, it typically requires applications to be shut down temporarily and introduces potential support challenges as with most custom solutions.
NAS-based data replication
Data replication between NAS devices offers by far the most advanced features and the best integration with applications. Vendors use a combination of local and remote replication as well as application-specific modules to provide NAS data protection. Examples of this technology include the NetApp "Snap" suite and EMC Celerra Replicator.
The NetApp data protection capabilities include snapshots for local, block-level, point-in-time copies, SnapVault for local or remote block-level incremental backups and SnapMirror for local or remote data replication. The suite also includes the SnapManager data management capability to automate and simplify the backup of application data such as MS SQL, Exchange, Oracle, etc., which require application-aware backups.
EMC Celerra leverages its Replicator and Replication Manager technology to provide asynchronous data replication over IP networks. It allows the creation of local and remote point-in-time copies and includes support for applications such as MS Exchange and SQL.
Both vendors have also integrated data deduplication technology with their solutions to help reduce the footprint of backup data.
Online data backup
Some NAS device vendors, focusing mostly on the small- to medium-sized business (SMB) market have developed online data backup offerings that are integrated with their products and leverage cloud storage services. Examples of that technology include CTERA Network's CloudPlug and Netgear's ReadyNAS Vault. Their main differentiator is the ability to provide a simple backup and disaster recovery solution to their clients.
It is clear that there are multiple options to backup NAS devices ranging from the traditional client based backup to the fully automated snapshot and replication offerings. Each option comes with a price but can also meet specific recovery requirements. Obviously, highly automated, block-level replication technologies will support much tighter recovery time objectives (RTOs) and recovery point objectives (RPOs) than a client-based, network backup to tape solution could. As always, the solution and associated cost should be justified by the recovery requirements.
About this author: Pierre Dorion is the data center practice director and a senior consultant with Long View Systems Inc. in Phoenix, Ariz., specializing in the areas of business continuity and DR planning services and corporate data protection.
This was first published in May 2010