Improving Backup Performance

When it comes to backup performance, every little detail counts. Unfortunately, some companies tend to oversimplify the discussion, focusing only on a handful of key factors. The reality is much more complex, and ignoring the finer points can put data protection at risk.

In a previous article (“Efficient Backup: Ready for the Black Scenario“), we explored backup methods, the importance of RTO and RPO, and the differences between data storage and backup. But that was just the tip of the iceberg. Now, let’s dive deeper and examine how storage media, network throughput, and data reduction mechanisms impact backup performance. What should you consider when choosing the best solutions for your business?

Storage Media: A Tough Choice

Not long ago, using SSDs in NAS systems (often used for backups) seemed unnecessary. Today, almost every NAS device supports them. One of the loudest voices advocating for ditching traditional hard drives is VAST Data, a U.S.-based company specializing in large-scale data management. Their solution relies entirely on flash storage, offering archive-level data retention at a cost comparable to HDDs. With proprietary technology that extends SSD lifespan to up to 10 years, their systems reduce the need for multiple devices to store and protect data. Perhaps in the future, businesses won’t have to choose between HDDs and SSDs – but for now, that decision still matters.

Flash storage significantly boosts backup performance. For example, a standard SATA SSD can be twice as fast as a mechanical SATA hard drive, while NVMe SSDs can outperform HDDs by a staggering 35 times. That’s a game-changer. Yet, many businesses still favor HDDs due to their lower cost. At the start of 2024, a 1TB NVMe PCIe 5 SSD cost around $150 – the same price as an 8TB HDD.

With the growing demand for storage-driven by explosive data growth and backup best practices (3-2-1, 4-3-2, 3-2-1-1-0) – choosing the right media is critical. Many experts argue that hard drives remain more reliable and predictable, while SSDs, despite improvements, can still experience unexpected failures.

But HDDs and SSDs aren’t the only options. Tape storage remains in play, offering a much cheaper alternative to HDDs for long-term archival. However, HDDs are far superior for short-term storage and incremental backups. Speed is also a factor – HDDs provide faster data access compared to tape, where retrieving data can take minutes due to loading and rewinding times. Restoring a large file system from 30 tapes could add up to two hours of delay, whereas HDDs work instantly.

Read more about Data Storage:

• Optimizing Data Storage Performance in Hybrid Cloud Environments
• Secondary Storage: Definition, Devices, and How It Can Support Your Backup Strategy

Network Throughput: The Silent Bottleneck

Backup performance isn’t just about storage; network speed plays a huge role, too. If backups run multiple times a day, high bandwidth is essential. But if backups occur less frequently – say, once a week or only after hours – the network load is lower. To pinpoint bottlenecks, companies should analyze their network’s data transfer speed. A simple formula can help:

Backup Data Size / Backup Window (time allocated for backup without disrupting operations)

For instance, backing up 5TB of data within a six-hour window requires a network capable of transferring 853GB per hour (5,120GB / 6 hours). On a Fast Ethernet (100Base-T) network, transferring 250GB takes about an hour, whereas a Gigabit Ethernet (1000Base-T) network is ten times faster. In this case, Fast Ethernet won’t cut it.

To improve network performance, businesses can:

• Extend the backup window
• Use dedicated high-speed networks
• Upgrade to Gigabit Ethernet
• Optimize data through compression and deduplication

Slimming Down Data: Deduplication & Compression

With data storage costs on the rise, companies are turning to compression and deduplication to reduce backup sizes.

Compression comes in two types:

• Lossy: Removes unnecessary data (e.g., reducing image quality slightly)
• Lossless: Keeps all data intact, essential for text files, executables, and spreadsheets

For example, a 50MB file with a 2:1 compression ratio shrinks to 25MB. However, compression can reduce network throughput while processing data. While a small quality loss in compressed images may go unnoticed, it matters for high-resolution projects.

Deduplication, on the other hand, prevents redundant data from being stored multiple times. It scans, divides data into blocks, and saves only unique ones.

A key metric here is the deduplication factor, which is often misunderstood. If a solution advertises a 10:1 deduplication factor, it doesn’t mean twice the reduction compared to 20:1. Instead, the percentage of data saved follows this formula:

% Data Reduction = 1 – (1/DD) x 100%

• 10:1 Deduplication: 90% data reduction
• 20:1 Deduplication: 95% data reduction

The difference between a 10:1 and 20:1 factor is only 5%, not double the savings as some assume.

Performance Optimization: It’s All About Balance

Reducing backup sizes is a smart move, but it must align with broader data protection strategies. Think of it like fuel efficiency in a car: a vehicle might have great mileage, but if the driver has bad habits – like underinflated tires or carrying unnecessary weight – it will still waste fuel.

The same principle applies to backup performance. Every component – from media type and network speed to compression and deduplication – affects the overall efficiency. The key is balancing cost, speed, and reliability to ensure optimal data protection.

By understanding these details, businesses can make smarter backup decisions—without sacrificing performance or security.