Reading Time: 4 minutes

There's no denying that your data, especially in unstructured form, is one of your organization's most valuable assets.

Losing access to data for extended periods of time can be crippling. The human and financial costs of data loss – whether temporary or permanent – to an organization continue to grow with every passing year.

Data availability has long been considered a storage problem, but it's not. It's a data management problem.

As the leader in global unstructured data management, our customers look to Panzura to maintain both the integrity of their data, and the availability of their data. Both are critical, in equal measures.

Layer One – Immutable Data

Panzura operates in the world of object storage – every nibble (half a byte) that our customers write into their Panzura environment is inherently immutable. That is, once an object is created, it cannot be modified.

Where we level up is in how we use this immutability.

In typical backup/replication systems, copies of your files and changes to your files are backed-up in their entirety. That doubles your storage consumption, and gives you a heavy dataset to restore from. It also means restores take time; usually, more time than you can afford.

By contrast, Panzura CloudFS generates metadata pointers that reference the immutable objects that make up a file, and updates them in real time, every time a file changes. Read-only local and file system snapshots then capture the metadata pointers – and therefore the contents of the file system – at any given point in time.

When our customers need to recover files, they just restore them from a snapshot taken prior to the update (or the damage) they need to undo.

It's quick, and it's easy because unlike restoring from a back up, our customers aren't restoring data, but lightweight metadata.  It's a fraction of the size of a file, so it takes next to no time to restore.

Additionally, because of the proprietary global deduplication process we perform, every data object is unique. If more than one file uses the same data, metadata pointers reference the object, rather than storing a full copy of the file.  That means less storage space consumed, and faster file restores, without data loss.

Oh, and we don’t charge our customers for snapshots!

Layer Two – Dedicated High Availability

CloudFS is designed as an enterprise grade cloud file system with no single point of failure, and allows redundancy to be built in at every step.

If a Panzura node becomes unavailable at a location for any reason, our customers look to their high availability nodes to offer business continuity. These high availability nodes can be designated local – configured to automatically takeover one or more locations in the event of failure, or global – configured to automatically takeover the entire file system environment if the master node suffers an outage.

Each of our customers' environments are unique in terms of their requirements for HA redundancy, so each is configured to appropriately balance immediate availability against cost.

This is particularly important when you realize that the Panzura node at any location can be swapped out and brought online in a matter of minutes, even without high availability.

So, when availability is crucial everywhere, high availability is our customers' secret weapon.

Layer Three – Cloud Mirroring

Every byte that our customers write into their Panzura environment is stored in the object store of their choice – this includes data, metadata, and configuration information. 

This makes the object store a critical component. Though it is mostly providing a place to store data, and CloudFS runs all the intelligence and processes before data sync - global deduplication, data and sessions encryption, if the stored data itself is unavailable for a period of time, you have a problem.

While CloudFS provides redundancy by design, it doesn't itself store data. The object store is the one component used by our enterprise global file system that could be a single point of failure.

Though the object store providers have resiliencies and redundancies in place, object stores can and do suffer outages. While this doesn’t put a stop to the Panzura environment, the object store is your single authoritative data source, as well as what we use to make your data ransomware extortion-proof, so outages can be very disruptive. 

To boost data availability and protect our customers from cloud outages, we offer cloud mirroring.

Cloud mirroring offers our customers the ability to maintain an identical mirrored data set in a separate object store of their choice. As changes are made by users and processes at the edge, the new data these changes generate are written into both object stores, simultaneously. 

This keeps the two data sets mirrored in real time, and in case the primary object store is lost, the secondary automatically (no human intervention required) takes over both read and write operations. That means no interruptions to users or processes, business continuity in a storage outage, and no data loss.  

Additionally, the two object stores are isolated from each other, creating a natural air gap. In case the primary object store is corrupted, deleted or impacted in any way, the secondary store will function without any ill-effects from the primary.

Cloud mirroring can be deployed using any combination of object stores, to suit our customers' requirements: on-prem and public cloud, two regions of the same public cloud, or two independent cloud providers.

But, What if There Was Yet Another Layer?

Relentless about impact is one of our core principles, and when it comes to our customers, there's nothing more impactful than knowing data is safe.

While we've make data safe, resilient and available with three protective layers, what if there was a fourth?

A layer of defense preceding these – one that could be more proactive in slowing or stopping attacks?  One that lessened the impact of any attack by catching it before it can gain traction?

Watch this space!