Early adoption: NAS and traditional backup
The typical NAS deployment includes one or multiple NAS filers, which are backed up via traditional backup software (i.e., CommVault, EMC Corp. NetWorker, IBM Corp. Tivoli Storage Manager, Symantec Corp. Veritas NetBackup, etc.). The most basic approach includes a surrogate backup host with backup client software, which mounts NAS volumes via NFS/CIFS, and copies files for backup over the network. Another approach is to leverage NDMP, which allows backup software to orchestrate backup data movement directly from the NAS device to physical or virtual tape devices, thus skipping a step in the data movement process.
Issues with a typical NAS deployment include:
- I/O bottlenecking, often at the network level
- Long backup windows
- Performance impacts to users
- Number of tape drive mount points for large NDMP backups
- Native version control by backup software
- Centralized backup schedules and management
Full adoption: NAS and Native Copy technology
Most NAS architectures include a copy function to create and maintain multiple copies of changed files and efficiently do so by storing block-level changes in files only. This local copy function is typically complemented by replication and snapshot clones, and can provide another layer of protection for disaster recovery. Most of these implementations still include traditional backup technology, as customers rarely eliminate traditional backup from the NAS environment.
- Limited visibility into NAS copy functions (monitoring and reporting capabilities are immature)
- As environment grows, copy and replication scheduling may grow complex to manage
- There are limits to the number of NAS copies (and versions), (e.g., archive or long-term retention)
- Deleted data (how long do your copies exist, and can you restore deleted data?)
- Space consumption by "copies" of deleted data (because of policies to avoid the problem above)
- Eliminated bottlenecks associated with traditional data backup
- Efficient use of bandwidth and disk because of the block-level incremental nature of copies
- Easy to configure and deploy
- Backups of data can still be made from replicated volumes at a remote site using NDMP
- Reduced data movement for backup and offsite replication
The future: NAS, with VTL interfaces and data deduplication
This is where the story gets interesting. Here's a brief market recap:
Data Domain had one of the first data deduplication products, which provided a native file-system mount point for backup software storage. As the market ripened for VTL solutions, Data Domain built VTL interfaces in addition to NAS interfaces, and has capitalized heavily on the small-midsized business (SMB) market with a simple and viable alternative to tape infrastructure for core and remote sites.
NetApp has recently made major push into the backup market by incorporating native deduplication and VTL interfaces into its NearStore product line, which is specifically targeted for the backup storage infrastructure market. The NAS implementation provides a deduplicated disk storage device, a suitable target for regular backup images or a traditional NAS implementation with the benefits of deduplication. The NearStore VTL implementation (with deduplication coming soon) provides a VTL solution for backup software, and again serves as a suitable target for backup images. This solution also includes an innovative capability to perform deduplication "inline" or "post-process" depending on workload conditions.
NEC Corp. of America is coming to market with a innovative NAS architecture with its HydraStor, which provides a grid-based globally deduplicated storage solution for backup applications. NEC's architectural approach is completely unique. Most competitors (VTLs included) adopted appliance-based systems architecture, and while simple to deploy, it has resulted in major design challenges associated with deduplication processing and performance.
Technology convergence usually results in increased competition and better products, which will ultimately be good for users, but potentially bad for early adopters. The deduplication designs of NAS devices are evolving, and vendors have yet to prove which deduplication method, architecture and overall integration model (NAS vs. VTL), will become the future industry standard. In the meantime, keep a close eye on the NAS/deduplication leaders and newcomers, as the NAS meets backup story continues to evolve.
About the author: John Merryman is Services Director for Recovery Services at GlassHouse Technologies Inc., where he's responsible for service design and delivery worldwide. Merryman often serves as a subject matter expert in data protection, technology risk and information management-related matters, including speaking engagements and publications in leading industry forums.
This was first published in April 2008