Unstructured data and all-flash backup: The magic is in the recovery

Marcel Rebelo, Business Development Manager, Data Sciences Corporation.
Marcel Rebelo, Business Development Manager, Data Sciences Corporation.

Data is the new gold, oil or even sand (which can be turned into glass or silicon). So many different metaphors try to capture the value of data. However, we can agree that data is valuable. And when data recovery takes too long, that generally leads to real business impact.

Under normal circumstances, data recovery might mean retrieving a few files that may take a couple of minutes, but when companies need to recover or rehydrate large chunks of data, this can take days or weeks. This situation is unfortunately becoming all too common thanks to ransomware, explains Marcel Rebelo, Business Development Manager at Data Sciences Corporation: "When recovering from ransomware attacks, you no longer have to restore one or two files that have been deleted or corrupted. Customers have to restore terabytes and terabytes of data because they have to bring their whole IT environment back. With demand for data being what it is today, there is no longer the luxury of having days and days to restore terabytes of data."

The unstructured data recovery challenge

On average, ransomware-related recoveries can take 22 days – especially when the attacked enterprise does not have an appropriate backup solution for their various types of data.

The issue is particularly daunting when it comes to unstructured data. This type of data represents all the e-mails, images, videos, documents, scripts and more that don't fit neatly into databases and other forms of structured data storage. Unstructured data in today’s businesses provides essential information for digital operations and high-performance computing tasks.

Rebelo says this data needs a new generation of backup appliance: "The growth in data is really in the form of unstructured data. Our clients are generating hundreds of thousands of small files and objects, ranging from medical images to financial transactional records. Traditional purpose-built backup appliances (PBBA) aren't performant enough when clients need to restore this data type and then process it quickly, because of the disk latency when recovering the volume of data at scale."

Backup systems using magnetic disk hard drives organise information across the array. This distribution of data blocks takes much longer to recover large batches of data and the latency is particularly noticeable when recovering unstructured data, which isn't conducive to neat organisation. When customers restore terabytes after a ransomware attack, it can take several days or weeks and can cost as much or more than the requested ransom.

Therefore, the storing and recovering of large structured and unstructured data backups at high speed requires a new breed of backup appliance.

The magic of all-flash recovery

Flash storage has been available for over a decade and earns its reputation for being incredibly high performant compared to traditional magnetic disk. Flash technology has been adopted in data centres primarily for high-speed data transactions, but historically less so for backup purposes due to perceptions of higher costs. Yet flash technology is becoming more attractive as costs reduce and through the value gained from innovations such as space-saving deduplication and faster restoration speeds.

Since unstructured data, by its nature, cannot be as organised as structured data, it can become very unruly on standard disk drives. Comprising countless small files and objects, the excessive time it can take to recover large amounts of unstructured data from standard disk arrays becomes impractical.

"Today's flash architectures bring the cost of flash close to the cost of disks, and the restore speeds are starting to far outweigh any additional costs. Most of our clients now use all-flash arrays for their primary transactional and unstructured data and are now seriously considering flash as backup target storage.

“With ransomware, it’s no longer the old days where you restore one little file or a table in a database. You don’t know what’s been touched, and you have to get back to a point where you’ve got a clean copy of data before you can start moving forward again.”

Fast access to unstructured data

Flash storage works differently from standard disks, providing a solution to slow disk-based data recovery. A backup appliance using all-flash can achieve recovery times 50 times faster than traditional disk-based arrays. The performance is so impressive that some of Rebelo's customers use separate all-flash arrays for production and backups.

"Flash storage performance is so good that we're seeing more use cases appear outside of backups. We've got clients in the financial, medical and insurance industries who are using VAST Data appliances for unstructured data; some for backup, others for transactional file-type data or small transaction objects. They are not using one appliance for both as that would entirely defeat the purpose of making backups. But the fact that they will invest in two separate systems tells you a lot about how well flash-storage like VAST Data performs with unstructured data and backup data restoration."

Gartner and IDC both estimate that unstructured data represents at least 80% of all new enterprise data. It's growing three times faster due to use cases such as business intelligence, customer analytics, marketing intelligence and regulatory compliance. When disaster strikes, traditional backup appliances cannot quickly recover this data. But all-flash arrays are up to the task, showing the magic of high-speed recovery.

Share