Flash storage: the good and the bad

data centres with 0s and 1s overlaid

Flash storage can be potentially transformative for businesses, but it’s not all plain sailing…

The solid-state storage market is on the march. By 2022, it could be worth $25.3bn, growing almost 10% per year. The rise of cloud computing, falling costs, and performance advantages over HDD are all combining to generate a huge amount of growth.

Flash, it seems, is the future, but as so often it’s not quite that simple.

What’s good?

Flash arrays are faster. They deliver high performance, normally with a latency of around 1ms – ideal for applications needing to process a high volume of data every second. Sectors such as insurance, for example, are using digital technology to gather vast quantities of personal data, which help them deliver more personalised insurance products. Financial services need rapid data analytics to operate multiple trades within nanoseconds. These can feed into automated trading applications which deliver superior second-by-second access to the markets. All this data coming into their company helps improve their products and pricing, but it also means they need to dramatically scale up their data management capabilities.

Flash now offers greater data density than conventional storage solutions, so take up less floor space. What’s more, because they have no moving parts, they are more resilient.

What’s bad?

Of course, it is not all positive. There are issues about how flash arrays degrade. The NAND flash substrate can only go through a finite number of rewrites before it becomes unusable. When the information in a flash cell is erased, it is hit with a large charge of electrical energy. The semiconductor layer on the chip degrades a little each time, so whenever data is overwritten the system drops in performance.

Flash arrays tend to use a number of data reduction methods to manage deterioration, such as consolidating useful data on new blocks to reduce the damage done by re-writes, but this is resource-intensive, and can become less effective over time.

Cost remains an issue. Although the price has come down significantly in recent years, the per gigabyte cost is still approximately four times higher than HDD. Given that flash enables businesses to manage much higher levels of data, it also pushed up costs by increasing a company’s data management requirements.

What’s complicated?

There are many different types to choose from such as hybrid flash, all-flash arrays, and PCIe Cards. Greater functionality and flexibility inevitably brings with it greater complexity which in turn increases the administrative burden of data management.

An all-flash array provides higher performance, lower power bills and latency. On the other hand, the initial cost can be higher. The choice depends on many different factors, such as whether your organisation can afford the upfront cost and how it wants to use its data. Passive data, for example, which needs to be stored but not necessarily accessed regularly can be housed on old-fashioned traditional storage mediums.

The market is also evolving rapidly both in terms of the quality of technology on offer and the cost. As always, data managers will need to stay on top of these developments to offer the best advice and support.