A file system is a file system is a file system, right? Well, not really. While any short list would include FAT, NTFS, NFS, ZFS, XFS, OneFS, WAFL and, Nasuni UniFS®, all these file systems are not created equal.
Upon closer inspection, you see there are two fundamentally different design methodologies. The “old school” traditional file systems were designed assuming an underlying disk sector-block structure. Nasuni UniFS® was designed as a cloud-based, object storage global file system.
What are the technical differences and why does Nasuni’s new type of global file system matter? Let’s dig a little deeper.
Traditional File Systems: Maintaining the Status Quo
Most traditional file systems rely on storing data, both active data as well as snapshots, in sectors and blocks on disk. In a simple sense, metadata such as a table of contents, directory tables, inodes and other manifests keep track of which sectors and blocks belong to which file or snapshot.
This works fine until a failure or corruption event occurs, which is typically where RAID or other disk protection methods come into play. A drive fails and RAID rebuilds the blocks on the failed drive using parity or other methods to recreate the lost drive or data. This is a simple method that is proven and has been around for decades.
While this works for complete drive failure situations, it doesn’t address the corruption or deletion of a file, a series of blocks, or a directory table. Since disk blocks can be constantly overwritten, these forms of corruption are not uncommon. In these cases, you must rely on local backups, replication, or off-site tape to recover the lost or corrupted data. Restores, which rewrite the backed-up data over the disk blocks, will work most of the time, but usually at a significant cost of lost time and productivity. And then there are the times when they don’t work.
This is what most enterprises are faced with daily when managing the different types of unstructured file data across primary NAS, archive, and even snapshots. Such is the nature of a file system that provides read and write capability tied to a device.
Breaking With Traditional File Systems
How do you provide a file system with a master data set that is, for all practical purposes, incorruptible? How do you build a file system that can protect data without the usual added cost and complexity of backup, replication, or off-site tape storage?
These are two of the problems Nasuni set out to solve when the company was first started. Then, we added several more requirements to the drawing board that broke with traditional file systems:
- Unlimited scale
- No file size limits
- Unlimited retention for files and versions
- No limit on inodes or metadata to support billions of files.
- Global namespace to make data accessible in many locations simultaneously
- Global file locking to eliminate file conflicts
- Local caching to speed up end-user access and performance.
I could go on, but you get the idea. We were looking to turn the world of global file systems on its head.
To give my deeper dive on UniFS more meaning, a short history lesson is probably in order. Nasuni founder Andres Rodriguez and his team of founding engineers were also the brains behind Archivas, a private cloud, object-oriented storage company that was acquired by Hitachi Data Systems (HDS) in 2007 (Archivas now serves as the foundation for Hitachi Content Platform, or HCP).
Having created one of the first commercially successful object stores, Andres and his team knew its strengths and weaknesses better than anyone. While object stores are good for storing massive quantities of data, by themselves they offer limited accessibility for traditional applications, files, and protocols; limited data protection; and limited collaboration. How could Andres and his team write the next chapter, and pick up where Archivas left off?
They thought about building a gateway to pass files from on-premises storage to cloud object stores. But a cloud gateway wouldn’t solve the NAS scalability, data protection, and global file access challenges inherent in traditional file systems.
Instead, they decided to create the first global file system designed to “live” in cloud object stores, and unlock the full potential of what they had initially built at Archivas. That’s how Nasuni and Nasuni UniFS were born – Nasuni short for “NAS Unified” and UniFS short for “Unified File System.”
Nasuni UniFS: A New Global File System Approach
UniFS is a file system designed and built to live in an object store, not a traditional file system relying on an underlying, read-write disk sector-block subsystem. UniFS is designed around WORM principals and never over-writes an object once it is written. This means that files in the file system are kept immutable by UniFS. This holds true for file versions as well – every file change is time-stamped as its own object to provide complete data protection, eliminating the need for separate file backup and replication tools and processes.
Metadata is also stored in the object store, enabling UniFS to extend to any location using the table of contents, directory structure, inodes, and other manifests stored in the cloud.
When I talk about UniFS, this is where I start to get peppered with questions.
Q: But you have to get the data to the object store first, right?
A: Correct. Nasuni provides on-premises caching appliances for high performance access to files locally using traditional NAS protocols. File data created on these edge appliances is automatically propagated to the object store, where it is stored as the authoritative “gold copy” by UniFS.
Q: Can’t you have the same type of disk corruption on the local appliance that you have with traditional device-centric file systems?
A: Yes. However, the difference is the local copies of files on-premises are merely the “cached” versions of the UniFS gold master copies, which reside in the object store. With traditional disk-based file systems, the local copy is the gold master. Also, with UniFS, no data in the object store is ever overwritten. Both existing and new versions of files are written as WORM, which is why UniFS has never experienced a file corruption event. Traditional disk-based file systems cannot claim this, because they allow disk blocks to be overwritten.
Q: What does it take to recover file data?
A: All it takes to recover any file or set of files is to restore the metadata to a local edge appliance. Because the UniFS metadata is referencing previous uncorrupted objects residing in the object store, you’re able to recover any previous version of any file. This is true whether the file was deleted by a user, corrupted by an OS malfunction, or infected by WannaCry or CryptoLocker. This is also why so many customers use Nasuni as a fast Ransomware fix.
Q: How is Disaster Recovery handled by UniFS?
A: DR is as quick and easy as other types of recoveries. The metadata restore is lightweight and very fast, and can be performed from a central management console with a few clicks. You can quickly instantiate edge appliances at any site or location that has power and cloud connectivity and start populating them with the desired file versions within 15 minutes. The built-in geo-redundancy of the cloud combined with the global reach of UniFS means the days of building and maintaining expensive, dedicated disaster recovery sites are over.
A Next-generation Approach Equals a Better Outcome
There are many hardware and software solutions that can write data to the cloud. Most enterprise-grade, on-premises storage systems and data protection solutions have this capability. However, almost all of them start at the edge, use on-premises devices to store gold master copies of files, and push the files to the cloud for backup. Nasuni UniFS is the first cloud-native global file system that starts with the cloud, uses the object store as the gold master, and pushes files to the edge.
This approach is much harder to engineer. But, as explained above, it is the better approach for today’s modern, cloud-based infrastructures. Nasuni UniFS is the innovation behind a unified set of file services that offer unlimited capacity, improved recovery points and recovery times, global file access and collaboration, DR without the need for dedicated DR sites, and simple management.
The next time you think about moving your files to the cloud or upgrading to a global file system that was designed to scale with cloud object storage, think Nasuni and UniFS.
Learn more by downloading the white paper “UniFS® – A True Global File System” here »