AnandTech Storage Bench 2013 Preview - 'The Destroyer'

When I built the AnandTech Heavy and Light Storage Bench suites in 2011 I did so because we didn't have any good tools at the time that would begin to stress a drive's garbage collection routines. Once all blocks have a sufficient number of used pages, all further writes will inevitably trigger some sort of garbage collection/block recycling algorithm. Our Heavy 2011 test in particular was designed to do just this. By hitting the test SSD with a large enough and write intensive enough workload, we could ensure that some amount of GC would happen.

There were a couple of issues with our 2011 tests that I've been wanting to rectify however. First off, all of our 2011 tests were built using Windows 7 x64 pre-SP1, which meant there were potentially some 4K alignment issues that wouldn't exist had we built the trace on a system with SP1. This didn't really impact most SSDs but it proved to be a problem with some hard drives. Secondly, and more recently, I've shifted focus from simply triggering GC routines to really looking at worst case scenario performance after prolonged random IO. For years I'd felt the negative impacts of inconsistent IO performance with all SSDs, but until the S3700 showed up I didn't think to actually measure and visualize IO consistency. The problem with our IO consistency tests are they are very focused on 4KB random writes at high queue depths and full LBA spans, not exactly a real world client usage model. The aspects of SSD architecture that those tests stress however are very important, and none of our existing tests were doing a good job of quantifying that.

I needed an updated heavy test, one that dealt with an even larger set of data and one that somehow incorporated IO consistency into its metrics. The new benchmark doesn't have a name, I've just been calling it The Destroyer (although AnandTech Storage Bench 2013 is likely a better fit for PR reasons).

Everything about this new test is bigger and better. The test platform moves to Windows 8 Pro x64. The workload is far more realistic. Just as before, this is an application trace based test - I record all IO requests made to a test system, then play them back on the drive I'm measuring and run statistical analysis on the drive's responses.

Imitating most modern benchmarks I crafted the Destroyer out of a series of scenarios. For this benchmark I focused heavily on Photo editing, Gaming, Virtualization, General Productivity, Video Playback and Application Development. Rough descriptions of the various scenarios are in the table below:

AnandTech Storage Bench 2013 Preview - The Destroyer
Workload Description Applications Used
Photo Sync/Editing Import images, edit, export Adobe Photoshop CS6, Adobe Lightroom 4, Dropbox
Gaming Download/install games, play games Steam, Deus Ex, Skyrim, Starcraft 2, BioShock Infinite
Virtualization Run/manage VM, use general apps inside VM VirtualBox
General Productivity Browse the web, manage local email, copy files, encrypt/decrypt files, backup system, download content, virus/malware scan Chrome, IE10, Outlook, Windows 8, AxCrypt, uTorrent, AdAware
Video Playback Copy and watch movies Windows 8
Application Development Compile projects, check out code, download code samples Visual Studio 2012

While some tasks remained independent, many were stitched together (e.g. system backups would take place while other scenarios were taking place). The overall stats give some justification to what I've been calling this test internally:

AnandTech Storage Bench 2013 Preview - The Destroyer, Specs
  The Destroyer (2013) Heavy 2011
Reads 38.83 million 2.17 million
Writes 10.98 million 1.78 million
Total IO Operations 49.8 million 3.99 million
Total GB Read 1583.02 GB 48.63 GB
Total GB Written 875.62 GB 106.32 GB
Average Queue Depth ~5.5 ~4.6
Focus Worst case multitasking, IO consistency Peak IO, basic GC routines

SSDs have grown in their performance abilities over the years, so I wanted a new test that could really push high queue depths at times. The average queue depth is still realistic for a client workload, but the Destroyer has some very demanding peaks. When I first introduced the Heavy 2011 test, some drives would take multiple hours to complete it - today most high performance SSDs can finish the test in under 90 minutes. The Destroyer? So far the fastest I've seen it go is 10 hours. Most high performance I've tested seem to need around 12 - 13 hours per run, with mainstream drives taking closer to 24 hours. The read/write balance is also a lot more realistic than in the Heavy 2011 test. Back in 2011 I just needed something that had a ton of writes so I could start separating the good from the bad. Now that the drives have matured, I felt a test that was a bit more balanced would be a better idea.

Despite the balance recalibration, there's just a ton of data moving around in this test. Ultimately the sheer volume of data here and the fact that there's a good amount of random IO courtesy of all of the multitasking (e.g. background VM work, background photo exports/syncs, etc...) makes the Destroyer do a far better job of giving credit for performance consistency than the old Heavy 2011 test. Both tests are valid, they just stress/showcase different things. As the days of begging for better random IO performance and basic GC intelligence are over, I wanted a test that would give me a bit more of what I'm interested in these days. As I mentioned in the S3700 review - having good worst case IO performance and consistency matters just as much to client users as it does to enterprise users.

Given the sheer amount of time it takes to run through the Destroyer, and the fact that the test was only completed recently, I don't have many results to share. I'll be populating this database over the coming weeks/months. I'm still hunting for any issues/weirdness with the test so I'm not ready to remove the "Preview" label from it just yet. But the results thus far are very telling.

I'm reporting two primary metrics with the Destroyer: average data rate in MB/s and average service time in microseconds. The former gives you an idea of the throughput of the drive during the time that it was running the Destroyer workload. This can be a very good indication of overall performance. What average data rate doesn't do a good job of is taking into account response time of very bursty (read: high queue depth) IO. By reporting average service time we heavily weigh latency for queued IOs. You'll note that this is a metric I've been reporting in our enterprise benchmarks for a while now. With the client tests maturing, the time was right for a little convergence.

AnandTech Storage Bench 2013 -

Now we see what Seagate's balance of consistency and peak performance gives us: leading performance in our latest benchmark. The Destroyer does a good job of penalizing drives with poor IO consistency as the entire drive is written to more than once, but the workload is more client-like than a pure 4KB random write. The result is a test that likes both peak performance and consistent behavior. Seagate's 600 delivers both. I purposefully didn't include a 120GB Seagate 600 here. This test was really optimized for 400GB+ capacities, at lower capacities (especially on drives that don't behave well in a full state) the performance dropoff can be significant. I'm not too eager to include 240/256GB drives here either but I kept some of the original numbers I launched the test with.

OCZ's Vector actually does incredibly well here, giving us more insight into the balance of peak performance/IO consistency needed to do well in our latest test.

AnandTech Storage Bench 2013 -

The 600's average service time throughout this test is very good. OCZ's Vector does even better here, outperforming the 480GB 600 and falling short of the 400GB 600 Pro. The difference between the 600 and 600 Pro here gives you a good idea of how much performance can scale if you leave some spare area on the drive.

Performance Consistency Random & Sequential Performance


View All Comments

  • Kristian Vättö - Tuesday, May 7, 2013 - link

    The units we have are all based on the older 24nm NAND. A while back I asked Corsair for review samples of the 128/256GB Neutrons (the original ones are 120/240) but they said they are not sampling them (yet). I can ask if they have changed their mind, although there shouldn't be much difference since 19nm Toshiba NAND has the same page/block/die size as 24nm. Reply
  • FunBunny2 - Tuesday, May 7, 2013 - link

    Does "Toshiba" mean toggle-mode NAND, by definition? Or do they sell all types? Reply
  • Kristian Vättö - Wednesday, May 8, 2013 - link

    Yes, Toshiba uses Toggle-Mode interface for their NAND. Here's the breakdown of NAND interfaces and manufacturers:

    Toggle-Mode: Toshiba/SanDisk (joint-venture) & Samsung
    ONFI: Intel/Micron (aka IMFT, also a joint-venture) & Hynix
  • LtGoonRush - Tuesday, May 7, 2013 - link

    HardOCP showed pretty significant performance increases, though that could also be due to the new firmware (which is not being back-ported as I understand). Reply
  • romrunning - Tuesday, May 7, 2013 - link

    I really wish we had more tests of SSDs in RAID-5 arrays. This is really useful for SMBs who may not want/afford a SAN. I'm very curious to see if the 20% spare area affects SSDs just as much when they're RAIDed together as it does standalone. I also don't care of the SSDs are branded as being "enterprise" drives. It would be nice to see how a 5x256GB Samsung 840 Pro RAID-5 array would peform, or even a 5x400GB Seagate 600 Pro RAID-5 array. Reply
  • FunBunny2 - Tuesday, May 7, 2013 - link

    No legitimate RDBMS vendor would allow its database on a RAID-5 machine. Never. Never. Never. Reply
  • romrunning - Wednesday, May 8, 2013 - link

    I can't tell if you're just trolling or you're actually serious. Obviously, SMBs use RAID-5 arrays ALL the time, and they use "legitimate" database products like MS-SQL, etc. It doesn't have to be an IBM AIX server running DB2, or anything high-end. Reply
  • daniel_mayes - Wednesday, May 8, 2013 - link

    What is FunBunny2 talking about? What Raid would you want to run them on 1,5,6,10, no ssd's?
    You aren't the only one that want's to see more tests with SSD's in a Raid 5. I would also like to see the destroyer run on ssd's with a higher provision and please add Intel DC S3700 to the destroyer benchmark next.
  • FunBunny2 - Wednesday, May 8, 2013 - link

    "I always have found that based on those requirements RAID 5 requires more spindles to satisfy those requirements than RAID 10 - and this has been found even with a Read/Write of 9:1. "

    (no, that's not me)

    Fact is, SSD still writes slower than reads, so what kind of RAID one uses matters. Having a 3NF (or higher) schema is a more productive avenue for performance on SSD, anyways, irregardless. Getting rid of all that bloated [un|de]normalized byte pile will allow, in most cases, you to have a much smaller database, and thus not worry about bunches and bunches of discs.
  • romrunning - Friday, May 10, 2013 - link

    That blog is from 2007, and SSDs weren't really in the picture at all. It has been demonstrated how SSDs can trump spinning disks in virtually all I/O-bound operations. The man in the blog even showed a test of RAID-5 beating RAID-10 on the same hardware, so his test was in direct contradiction to the one who later commented about spindles.

    That being said, I think you're trying to say that getting rid of unnecessary in your database will result in a smaller database & thus lower performance requirements. That might be true at one point, but when you've normalized your data already, then additional data will just make the database grow. After all, if you're writing something like electronic orders to your normalized database, it will grow based upon real data addition. That's why you need to make sure your storage array can handle the increased load.

    RAID-5 has been the best for SMBs because it provide the fault-tolerance and the higher utilization of total storage capacity that they want. That's why I would like to see tests of SSDs in RAID-5 arrays - to get Anandtech to test these great SSD performers in something I could use in a database server. Something like their tests of their own website databases would be nice, or even smaller ones using a 10-20GB database.

Log in

Don't have an account? Sign up now