Solid state storage

For many companies, the time to adopt solid state drives will be when the extra speed and performance of solid state over traditional hard disk drives have an impact on a company's bottom line by bringing in more revenue. Solid state drives have been touted as the most important PC innovation in years. But because SSDs seem expensive at first glance, some have doubted their corporate computing value, value that becomes clear when you take a close look at total cost of ownership (TCO).

First, a word about storage space in corporate notebooks. A 2007 study conducted by McKinsey & Company found that approximately 70% of business professionals only need a 60GB or smaller hard disk drive in their notebooks. With 128GB SSDs now available, a growing number of IT managers are considering the technology for employees that are on the road or work from home occasionally, and PC suppliers are responding by offering more notebooks with an SSD option. SSD-equipped servers and desktops will follow.

In fact, since SSDs have no moving parts, are extremely durable, emit virtually no heat and offer better performance than most rotating disk hard drives, they are rapidly gaining supporters worldwide.

Even greater interest can be expected once corporate IT execs do a thorough SSD TCO analysis, where TCO is defined as the cost incurred over the entire PC product life cycle, taking into account not only acquisition costs but also costs associated with maintenance and downtime.

TCO

At roughly 10 times the cost of a hard drive, the price tag of an SSD will make many IT decision-makers hesitate. But there are a number of areas where use of SSDs will lead to savings, both in network PCs and, eventually, network servers.

The largest area of savings can be attributed to the exceptional reliability of SSDs. The same McKinsey study estimated that 44% of firms experience a notebook hard drive failure rate of 5% to 10% annually, while another 37% suffer more than 10% notebook PC hard drive failures per year.

The downtime attributed to these failures can be expensive in terms of business processes idled, but the costs should also include the time and output gains lost when each employee with a faulty PC is unable to work. For example, a consultant or lawyer who bills hundreds of dollars per hour would recoup the cost of an SSD in a few hours. Further, there can be significant labor costs associated with the IT resources used to replace damaged drives and recover lost data. Additionally, SSD is immune to fragmentation, so additional performance degradation and associated de-frag delays can be avoided.

Performance gains

Beyond reliability gains, SSDs also bring significant performance benefits. For example, boot time on notebook PCs is about twice as fast, and applications launch 1.5 to 3 times more quickly. Read and write performance can be up to five times faster. For servers and storage, the benefit depends on the application, but performance gains average about three times more input/output operations per second, especially for random read tasks, as compared with a high-end hard disk drive.

Due to these performance gains, one SSD has the potential to replace four or more hard drives, reducing cost while at the same time increasing server productivity. And it should be noted that the mean time between failure for an SSD is 2 million hours versus 300,000 hours for a typical hard disk drive.

Power consumption is also a factor that IT decision-makers should consider. SSDs use 30% less power in notebooks than hard disks, which extends battery life in laptops about 10% and offers potentially dramatic electricity savings when used in enterprise servers. Moreover, SSDs don’t generate heat like hard disks, relieving costs associated with cooling.

Flash storage gets enterprise attention

The little USB stick on your keychain and the memory in your iPod is fueling a revolution in the enterprise storage world. It seems everyone is talking about flash memory, a type of solid-state storage that offers faster and more energy-efficient performance than rotating disk drives. The downside is that it's about 20 times more expensive than high-performance Fibre Channel drives, but that's where the popularity of USB sticks and the iPod comes in. The consumer demand for flash and another major event — EMC's entry into the enterprise flash market this year — are combining to drive prices down, making it feasible for enterprise use, experts say. Big businesses are already starting to use flash storage for I/O-intensive applications, such as Oracle databases, credit card processing systems and stock trading applications. Many observers expect solid-state flash drives to be commonplace in enterprises within a year or two.

“With data growth expected to reach 10-fold increase over the coming 5 years, it's important that organizations address that growth with efficient solutions. Cost reduction and reduction in power consumption as well as achieving faster results and reliability come as the main drivers behind Flash based storage adoption. Flash drives use 97.7 percent less energy per IOP than 15,000 RPM Fiber Channel disk drives,” says Said Akar, Pre-Sales Manager, EMC Middle East & North West Africa

David Beck, Storage Group Manager, Sun Microsystems MENA, offers a different perspective: “Actually, the need to have faster and higher performing applications, databases and particularly complex OLAP queries. By creating Tier 0 storage and enabling very frequently accessed files and database tables (sometimes the entire database) to reside there, performance can be improved dramatically. Some customers are also looking for the energy savings that Flash technology brings over and above traditional Hard Drives.”

That begs the question of whether flash storage will replace traditional hard disk drives. “Flash is much cheaper than DRAM but its slower, so by combining some of each we can achieve the right blend of price and performance. Using these blends we have created both “Write Optimised” flash and “Read Optimised” flash and we use them together in our systems and storage products. Will Flash replace HDDs? Maybe one day yes when prices come down significantly, but for now there's still a place for capacity oriented disk drives and performance oriented Flash in a enterprise architecture – Tier 0 and Tier 1,” says Beck.

Akar says flash is not a replacement of DRAM. “Where DRAM can be needed as an extension of server memory because if offers low latency for reads and writes, Flash is complementary and better suited for data storage and provides faster retrieval of data from storage subsystems and has significantly less latency time than traditional mechanical discs. So eventually Flash will become mainstream and will replace traditional hard drives but in a few years time.”

Solid-state technology, including flash memory, is certainly on the upswing, with IDC predicting 76% annual shipment growth through 2012 in a market that generated nearly $400 million in revenue in 2007. Fowler predicts that the majority of organizations building I/O-intensive applications will use some form of flash within a year.