Changing Of The Guard: 2.5” Hard Drives In The Enterprise

2.5” Vs. 3.5”: Capacities And Cost

There are a few things to consider before talking about capacities and what they actually mean. In general, hard drive makers aim to create drives with a reasonable number of rotating platters. Single-platter drives are most interesting in the consumer and client markets, where minimum cost is imperative.

Multiple platters are required to reach higher capacities or to reach certain capacities based on a proven technology and drive density. However, fast 3.5” drives with many platters try to combine high performance with high capacity, often entailing high cost. A 3.5”, 7,200 RPM drive will offer three times more capacity at one-third of the cost, while SSDs are conquering the performance segment.

Medium capacities are left, representing the enterprise mainstream. These can be managed best on the 2.5” form factor. Yes, there’s a small performance tradeoff, but power, efficiency, and cost balance well here. In addition, one product cycle is typically enough to make up for performance impacts. Existing storage densities have 10,000 RPM, 2.5” drives reaching 200GB per platter. This allows Seagate and Toshiba to create 300GB, 450GB, and 600GB capacity points. We expect Hitachi to follow soon.

The Capacity View

Given that many more 2.5” drives can be deployed than 3.5” drives into the same rack space, the smaller drives yield much greater storage density and power efficiency per gigabyte. Two 2.5”, 300GB, 10,000 RPM enterprise drives in a suitable RAID will outperform one 600GB, 3.5”, 15,000 RPM drive. At the same time, cost and power consumption are roughly comparable.

The Performance View

If we look at a 3.5” vs. 3.5” scenario, you typically need multiple drives to meet performance, capacity, or efficiency demands. In larger envorinments, storage deployments include not just individual drives but entire JBOD units. Let’s create a simple example.

A storage solution has to deliver a minimum of 1,000 file server IOPS and provide at least 3TB net capacity. The ideal target would be a 1U storage solution containing four 3.5” drives. With 600GB, 15,000 RPM units, we would achieve the required performance but fail in capacity. A 2U system would allow us to increase the drive count, but overall cost increases as well. The alternative is a 1U storage box that can hold ten 2.5” drives. Now it’s possible to install six 2.5”, 600GB, 10,000 RPM drives. Configured as a RAID 5, we’d have the required capacity and performance at a lower total cost, lower power consumption, and much higher power efficiency than the 3.5” solution.

To conclude, let’s imagine the cost difference if you were to use SSDs. A single drive could probably deliver the required peformance, but we’d need at least 24 drives at 128GB each to reach the net capacity. At this point, no redundancy would be included and the final solution would be massive. We’d have to plan the RAID array, check for suitable RAID controllers, and find storage appliances to operate 24+ drives. You get the point.