Zinho's Journal: Justify an SSD to your boss

I made a comment a while back detailing money I saved switching to a solid state drive (SSD) from a standard spinning-rust hard drive (HDD). Since then I've gotten a request to detail my procedure. It's not difficult or complicated, so I'll explain it here.

Step 1: profile daily usage. I work in a corporate environment, and my highest frequency applications are MS Outlook and SolidWorks; as a result, I took a stopwatch and timed a cold boot, launching Outlook (no document) and launching Solidworks (no document). I also timed how long it takes to open a file once solidworks is up. The times I got were 270s cold boot (power on to login prompt), 50s SolidWorks, 30s Outlook start, 15-20s document open. Yes, these are horrible. No, it's neither comprehensive nor precise; I was going for representative and ballpark figures.

Step 2: Find difference in use times on an SSD. My first draft approached this with a throughput estimate - seek time is minimal for most files compared to read time in my use case. The HDD that came with the system is specced for 100 MB/s max transfer rate, the SSD I wanted to buy (Crucial RealSSD C300) claimed up to 355MB/s max sequential read; from this I estimated about 1/3 the time to open a given file. Based on this I was able to run the rest of the "time=money" numbers and convince my boss to buy one as a trial. If you can get actual numbers instead of using estimates that's better; more on that later.

Step 3: Estimate time saved between HDD and SSD. I estimated one bootup and 20 solidworks files opened per day; I could have gone further, but the ROI numbers just on that were good enough. With the 1/3 assumption we're looking at 6 minutes a day wasted waiting for a disk to spin.

Step 4: Estimate cost of lost time. I tried to keep this from being personal by using my company's pay scale for a typical full-time employe at my level instead of my own salary. Dividing by work days per year, hours per work day, etc I was able to figure out a cost per minute of salary time. Multiplying that by 6 minutes/day gave me about $4/day.

Step 5: Calculate ROI time for new SSD. Knowing the cost of the drive you can then estimate how long it will take to pay itself off. For a $300 drive that is $300/$4 = 75 working days, or 15 calendar weeks (5 work days per week). That's less than 4 months. In fact, for an employee who has two weeks of vacation/holidays per year (250 work days) the drive earns back about $1000/year. Not bad.

Step 6: Get actual hard drive and refine the numbers. Once I had the sample drive in hand I re-profiled the same tasks as before. The new numbers were 75s cold boot, 15s SolidWorks start, 4s Outlook start, and 3-7s SW drawing open; this validated the 1/3 estimate, and allowed me to make a final report with real results. The final numbers weren't much different, though; 195s saved per bootup is not a lot more than 180s saved, and 12s/file isn't a big gain over 10s. We're still looking at about $4/day savings (ballpark). It was nice to be able to show that the real numbers lined up with the rough estimate (better than the estimate, actually).

Please note that to adapt this to yourself you'd need to decide where your time is spent waiting on disk for your profession; programmers, game developers, legal assistants, and graphic artists would presumably all have different applications to load and different amounts of disk wait time in their day. Total savings also are dependent on salary; a $150k/year salaried employee's time is much more costly than a $15k/year hourly minimum wage employee. Your management may be comfortable with a larger drive based on these numbers as well, that's up to your manager and where he/she reaches sticker shock. Also, try to be honest about the nature of this time saving. Large blocks of disk wait can be compensated for by doing other productive activities (even if it's only getting coffee during bootup instead of watching compy boot while sipping the morning coffee).

In my opinion, though, the biggest benefits are intangible. By not interrupting my work cycle for 15s blocks while I'm immersed in my thought process I improve my ability to concentrate on the task at hand. Having programs start immediately when called upon greatly improves the computing experience as well, and improves morale (the computer is working for me rather than fighting me or slowing me down). The experience is so much better that I'd recommend it to anyone working at a computer as their primary job.

As time passes the brand of drive you'll want to buy will change. These days I'd still recommend the C300's successor, the Crucial M4 256GB (~$200 on Newegg); however, the Corsair Force 3 and Force GT series are good bang-for-buck solutions as well (~$210 on Newegg for 240GB, better performance). I use techreport.com's occasional reports to keep an eye on the lay of the land as far as SSD value across the market. Other attempts I've seen didn't have the same numerical rigor to them or as eye-catching of a graph.

To anyone using this as a pattern to convince their boss that SSDs are worth it, good luck! It's been a great switch for me, and I hope the experience goes as well for you.