Go to ...

RSS Feed

August 11, 2020

Interesting US$200 2.5″ Storage Experiment


I’ll admit it: I’m a hardware junkie, particularly for storage stuff. Thus, when I learned that 5TiB (!) 2.5″ drives were now available, I had to have one. And, because a drive without an enclosure isn’t terribly usable, and these big drives are 15mm tall, I had to have a new enclosure to house that new drive, too. Here’s what I’m talking about:

+1: Seagate 5TiB BarraCuda 5400 RPM 128 MB: US$156.95 (including S&H plus sales tax)
+2: StarTech S251BU31315 2.5″ USB-C/3.1 Enclosure: US$40.51 (including S&H plus sales tax)

These items are depicted side-by-side in the lead-in graphic for this story, and come with a combined cost of US$197.46 (I rounded up for the story title).

Why Make This Purchase?

Given that 2 TB 7200 RPM drives currently sell for around $100, why would I want to spend another US$60 more to up the storage total to 5 TiB instead, while possibly losing out on performance? Good question! As a hardware junkie, I have to admit the purchase came as much from curiosity as from any need for additional storage devices. That storage inventory is all over the place, in fact. I’m using 8 3.5″ drives with a combined capacity of 31 TiB right now, with another 3 TiB 3.5″ drive on my bookshelf. I’ve got over a dozen other 2.5″ drives, of which half are SSDs ranging in capacity from 120GiB to 500 GiB, and the rest are HDDs at 1 or 2 TiB capacities. I’ve got 4 mSATA SSDs in USB drive enclosures with capacities at 0.25, 0.5 and 1.0 TiB, and one M.2 NVMe drive enclosure with 1.0 TiB capacity. But I’ve never seen, let alone installed or used a 15mm 2.5″ drive before, and I was curious about performance vis-a-vis 7200 RPM and SSD devices. So I bought one to assemble, play with and learn from. But please: consider this a preliminary report because I haven’t had much time to learn from experience and usage just yet.

Assembly, Set-Up and Use

A total of 6 screws are needed to assemble the drive into its case. Four screws hold the drive itself onto the drive skid (that slides inside the enclosure). Two teeny-tiny screws secure the skid to the case’s outer shell.Took me less than 2 minutes to put the whole shebang together. The enclosure includes those screws, and one each 55 cm (~22″) USB-C and USB 3.1 cables. There’s even a small (plastic!) Phillips head screwdriver, in case you don’t have a small one handy (I keep set of flat and Phillips head jeweler’s screwdrivers around, for just this kind of thing). Once the drive and enclosure were mated, I plugged the USB 3.1 cable in and connected it to an open front case port on my production desktop. It came up immediately in the Disk Management utility, where I formatted it with a single GPT partition, and named it SeaBar5 (for Seagate drive, type BarraCuda, capacity 5 TiB).

Interesting US$200 2.5' Storage Experiment.diskmgmt

Here’s what the drive looks like, formatted, named and granted drive letter “T:”
[Click image for full-sized view.]

The View from CrystalDiskMark

I compared performance for this drive with two other points of reference. All used the same USB 3.0 port on my production PC, and both HDDs used the same cable from device to port. Point 1 was my Sabrent EC-NVMe enclosure with a 1 TiB Samsung 760 NVMe SSD. Point 2 was a Samsung Spinpoint 2TiB 7200 RPM drive inside an Intatek USB 3.0 drive enclosure that I’ve owned for a little over two years.

5TB Seagate 2TB Samsung Spinpoint
1TB Samsung NVMe 760
Read Write Read Write Read Write
SEQ1GiBQ1 149.03 143.69 130.85 134.56 284.51 291.30
SEQ1GiBQ8 148.33 143.31 129.80 128.73 280.07 286.26
RND4KQ32T16 0.56 6.17 0.51 0.96 36.09 84.89
RND4KQ1 0.49 5.75 0.41 0.89 31.82 57.27

Note: CrystalDiskMark tests from top to bottom: SEQ1GiBQ1 = Sequential access, 1 GiB data chunk size, queue depth = 1; SEQ1GiBQ8 = Sequential access, 1 GiB data chunk size, queue depth = 8; RND4KQ32T16 = Random access, 4K data chunk size, queue depth=32, thread count=16; RND4KQ1 = random access, 4K data chunk size, queue depth =1; all items with no thread count included means thread count = 1. These are the defaults for the version of CDM I used: 7.0.0 x64.

I went into the testing convinced that the newer 5400 RPM higher-capacity drive — namely, the Seagate BarraCuda 5TiB — would significantly lag behind both of the other drives, performance wise. I was wrong. The newer higher-capacity HDD outperforms the older lower-capacity HDD by about 10% (or better, for random 4K writes), even though the older, lower-capacity drive rotates at 7200 RPM to the BarraCuda’s 5400 RPM. As you’d expect, the SSD blows away both of the HDDs, with around double the performance (or better for random reads and writes).

What do these results show? For one thing, they show me the recent purchase was not a waste of money. They also show what everybody already knows: HDDs remain great for backups and for external file storage and archiving. You surely don’t want to use them as boot drives any more, nor for storing data files and applications where disk latency can impact performance. Bottom line: I feel better about my experiment in light of these results, and would recommend the drive/enclosure combo to others looking for relatively compact and portable, more capacious storage for their notebook, laptop or tablet PCs. On a price per GiB basis, in fact — ~US$40/GiB for the Seagate BarraCuda 5TB plus enclosure versus US$60/GiB for the Spinpoint 2TB plus enclosure (US$100 for the drive, US$20 for the enclosure) — the newer, slower drive is actually a significantly better deal. That’s what I like to see! I plan to use it for backups, external storage and archiving, precisely the areas in which it will do me the most good.

Author: Ed Tittel

Ed Tittel is a 30-plus-year computer industry veteran. He’s a Princeton and multiple University of Texas graduate who’s worked in IT since 1981 when he started his first programming job. Over the past three decades he’s also worked as a manager, technical evangelist, consultant, trainer, and an expert witness. See his professional bio for all the details.

Tags:

2 Responses “Interesting US$200 2.5″ Storage Experiment”

  1. Toni Fasth
    June 24, 2020 at 19:38

    Interesting, seems your testrig does not support UASP. You should get closer to 1GB/s sequential R/W for an NVMe drive.

  2. July 18, 2020 at 00:50

    Tony, because of our recent website issues I haven’t been able to respond to your comment. Thanks for pointing this out to me. It actually spurred me to write a more recent post on June 26. It’s called UASP Outperforms BOT Significantly (find it at https://win10.guru/uasp-outperforms-bot-significantly/, if you haven’t seen it already). And indeed, I did get more than 1 Gbps from the NVMe in USB on the UASP capable Lenovo X380 I ran CrystalDiskMark on, just as you predicted. Thanks!
    –Ed–

Leave a Reply

More Stories From 20H1