Who Uses SSDs, And Why?

The arguments of whether to use solid-state disk (SSD) drives vary depending on the market. In certain markets, SSDs are a very reasonable choice, while in other markets SSDs are unlikely to gain much traction.

Objective Analysis divides the data processing market into four sectors: budget PC, notebook, desktop, and server—each with different storage requirements.

Budget PC

The budget PC category includes very low-cost systems, such as Asus’ Eee PC, Intel’s Classmate PC, etc., These systems are price-driven, so their storage requirements are dictated by price rather than capacity.

Industry-standard hard disk drives (HDDs) have an absolute minimum price (com- monly referred to as the “floor price”) of around $35 to $50—they just can’t be built to meet a lower price target. Meanwhile, a low-density NAND chip can be purchased for less than $1. The net result is that NAND flash can be used to build less costly storage systems, giving more flexibility to the overall bill of material (BOM) cost. Even though the price per gigabyte of NAND flash is more than 10 times that of an HDD, a cost-constrained system will often use $20 or less of flash for storage rather than a $50 HDD.

The figure above (from Objective Analysis’ report, The Solid-State Disk Market: A Rigorous Look) is a graphic depiction of this phenomenon. As capacity increases beyond an HDD’s floor price, HDDs offer the lower price for any given capacity (the right side of the vertical line). At capacities significantly lower than the capacity of a single HDD, flash costs less than HDDs.

Flash memory has been successful in budget PC systems, but the low price of flash memory and the small unit volumes sold in this market prevent the NAND market opportunity from growing very large in this category.

Notebook PCs

For several reasons, most NAND-based SSD manufacturers are focusing their marketing efforts on the notebook PC:

  • The market is large and growing rapidly, with more than 100 million units shipping every year.
  • Notebook PCs have higher BOM budgets than do budget PCs, so more flash can be sold into these systems.
  • Notebook PCs benefit from the inherent advantages of SSDs—high read speed, silent operation, low power consumption, and less sensitivity to shock, vibration, temperature, and altitude.

To date, however, consumers have been slow to warm up to the notion of spending an extra $500 or so for these advantages.

So far, Objective Analysis has been relatively negative on the prospects of SSDs in the notebook market, although a new campaign launched by San- Disk is altering our viewpoint to some degree. SanDisk has found that today’s economic slowdown has caused IT managers to push out the deployment of new notebooks, asking users to hang on to their existing notebooks for another year or two. Today’s three-year-old PC typically includes 256MB of DRAM and a 20GB HDD. An IT manager can upgrade users’ DRAM to 2GB and replace the 20GB HDD with a 32GB SSD, giving users a significantly improved system at an additional cost of only about $75 per notebook.

Although we do not know how large the market is for such upgrades, it is conceivable that millions of PCs could be retrofit with SSDs, boosting the SSD market significantly while saving costs in IT budgets.

Desktop PCs

Very few of the SSD attributes that are being pitched to notebook users are relevant to desktop PC users. Power is rarely a concern with desktops, and the systems are not sub- jected to shock and vibration. And it is very rare that desktop PCs are subjected to extreme temperatures or altitudes.

As for the SSD’s higher performance, it seems that only very serious gamers are willing to pay a few hundred dollars extra for a drive that improves response time. And the final SSD attribute, silent operation, has only been an issue in home media centers—a market that has not developed as rapidly as once anticipated.

Objective Analysis sees little reason for SSDs to have any meaningful penetration into the desktop PC market unless low- capacity SSDs are used by IT managers to postpone deployment of new systems, as in the notebook example.

Servers

Although SSDs are unlikely to fare well in the notebook and desktop PC markets, the server market is a different story. Objective Analysis expects the enterprise server space to be the highest growth opportunity for SSDs.

To understand why SSDs are attractive in the enterprise space, you have to understand how data center managers build storage systems. Data centers are assembled using a large number of HDDs. Some are chosen for their capacity, while others are selected for speed. This is where you typi- cally see enterprise HDDs with faster interfaces, such as Fibre Channel and SAS, and with spindle speeds of 10,000 or 15,000rpm, as opposed to the 5,000 to 7,500rpm common in PC-based HDDs.

These faster drives, which have lower capacities and higher prices than PC HDDs, are used to crank up the system’s speed, but at a price premium. In some cases, even these drives are not fast enough, and a variety of techniques are used to coax more speed out of the storage system, such as striping (spreading the data across several drives to increase I/O speed), RAID, and short-stroking (limiting head movement to speed access at the expense of lost storage capacity.)

The figure above illustrates the storage hierarchy as a pyramid, where the short-stroked drives (which we dub “abused HDDs”) appear at the top, since they are few in number and high in cost, while fully-used enterprise HDDs and high-capacity HDDs fill in the middle and lower levels of the pyramid.

Since they are few in number and high in cost, short-stroked drives are most susceptible of being replaced by SSDs.

Short-stroked drives are the category that is most susceptible to being replaced by SSDs. IT managers find that they can replace a number of HDDs with a single enterprise SSD at about the same cost and realize higher performance with a smaller footprint, lower power consumption, and less cooling. Objective Analysis expects this market to represent the biggest opportunity for SSDs in the near term. Although a very small number of HDDs fall into this category, the high price of enterprise HDDs leads us to predict that the enterprise SSD market will grow to more than $1 billion by 2013.

Two-pronged development

SSDs are well-positioned to penetrate the two extremes of computing—the very low end and the very high end—for very different reasons. At the low end, SSDs give OEMs a price point that is much lower than HDDs. At the high end, SSDs reduce drive count and the power, space, and cooling costs that come with it, while maintaining or even increasing performance levels. In both cases, SSDs save money.

Objective Analysis expects the SSD market to develop in these two markets first, before appreciable progress is made in the notebook and desktop PC markets.