Solidigm D7-P5810 800GB Enterprise SSD Review

Introduction and Drive Details

SK hynix is the second-largest memory company in the world. The South Korean company recently became so when it acquired Intel’s solid-state storage division and associated fab interests. SK hynix basically left in place what Intel had built and simply changed the name of its new subsidiary to Solidigm. We speculate that the driver of this massive acquisition was and continues to be a leg up for SK hynix as it relates to 4-bit flash technology and secondarily broadens its footprint in the enterprise space.

Why develop the world’s best QLC NAND technology when you can just buy it? So, what was Intel NAND is now Solidigm NAND. Indeed, as we’ve seen, ever since the introduction of SK hynix’s Solidigm brand, there has been a flood of game-changing QLC-based SSDs on the consumer side and many enterprise solutions.

4-bit flash, or QLC (Quad Level Cell) as it’s better known, is currently the densest form of commercially available flash, meaning the highest bit count in relation to physical device footprint. Capacity as it relates to physical footprint is arguably the most important factor in relation to TCO in the storage tier of the datacenter, where read frequently, program infrequently reigns supreme.

This is where Solidigm QLC SSDs are changing the game – read-centric workloads. 4-bit flash, for the most part, can serve data to the host (read) at a comparable rate to lower bit-count flash (TLC, MLC, SLC). However, it takes longer (higher latency) to program (write) than does lower bit count flash. Additionally, programming QLC requires on average significantly more wattage than lower bit-count flash. Finally, QLC flash is inherently less endurant than lower bit-count alternatives.

So, we can summarize in a rudimentary fashion that QLC in the datacenter is a balance of footprint vs. latency vs. endurance and that it is a balance that ultra-high capacity QLC SSDs are winning as epitomized by the world’s highest capacity commercially available datacenter SSD, Solidigm’s 61.44TB D5-P5336.

QLC SSDs, especially those with an overwhelming capacity proposition, are carving out a significant portion of the datacenter because in many read-centric roles they do, despite their inherent disadvantages, offer a better overall value proposition (TCO) than TLC-based arrays. But what if those inherent disadvantages can be easily augmented to the point where they all but disappear?

This is what we have for you in today’s review. Solidigm’s D7-P5810 is designed specifically to augment today’s massive QLC arrays as follows:

VIEW GALLERY – 35 IMAGES

So, the SLC-arrayed D7-P5810 is essentially designed to serve as a high endurance, low latency super-cache of sorts for the QLC mass storage array.

But wait, there’s more. Most datasets are random in nature. Random data is a nemesis of sorts for 4-bit storage arrays. Not only is it slower to write than sequential, but it is also much more taxing on the NAND than data that is sequential in nature. Well, Solidigm has an answer for that as well. Solidigm CSAL. CSAL stands for Cloud Storage Acceleration Layer. CSAL caches random writes and reshapes them into large IU aligned sequential writes as follows:

As we see it, CSAL is revolutionary. It literally erases any vestiges of the drawbacks inherent to QLC storage arrays. Ultra-density without penalty. Mind blowing.

Okay, now that we’ve briefly touched on what Solidigm’s D7-P5810 is all about, let’s take a close look at how well it performs, especially at low queue depths where it matters most.

Specs/Comparison Products

The D7-P5810 is currently available at 800GB capacity, with a planned 1.6TB option coming available beginning H1 2024. Endurance is massive at fifty 4K random drive writes per day (DWPD). Maximum power consumption is low at a mere 12 watts max. Targeted uses include write cache, HPC, write buffer, log journal, OLTP, or any other high-performance/low-capacity niche.

Solidigm D7-P5810 800GB NVMe PCIe Gen4 x4 U.2 15mm SSD

Enterprise Testing Methodology

TweakTown strictly adheres to industry-accepted Enterprise Solid State Storage testing procedures. Each test we perform repeats the same sequence of the following four steps:

1. Secure Erase SSD
2. Write the entire capacity of SSD a minimum of 2x with 128KB sequential write data, seamlessly transition to the next step
3. Precondition SSD at maximum QD measured (QD32 for SATA, QD256 for PCIe) with the test-specific workload for a sufficient amount of time to reach a constant steady-state, seamlessly transition to the next step
4. Run test-specific workload for 5 minutes at each measured Queue Depth, and record results

Benchmarks – Random and Sequential

4K Random Write/Read

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. Steady state is achieved at 2,000 seconds of preconditioning. The average steady-state write performance at QD256 is approximately 523K IOPS.

Solidigm specs its 800GB D7-P5810 as capable of delivering up to a massive 495K 4K random write IOPS. We can exceed this by almost 30K IOPS. In fact, only the Gen5 SSDs that appear on our chart can deliver more. Additionally, it is important to keep in mind that our test subject can deliver this with a minimum of 50% less power consumption, doing so with a massive 50 DWPD endurance on tap.

Low queue depth random read performance is performance that matters, and our test subject can deliver here better than any flash-based SSD we’ve ever tested. Not even the two Gen5 SSDs appearing on our chart can come close. This is supreme efficiency, especially considering the D7-P5810 is doing so with only half the power draw of our Gen5 contenders. Impressive.

8K Random Write/Read

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. Steady state is achieved at 1,000 seconds of preconditioning. The average steady-state write performance at QD256 is approximately 267K IOPS.

We expect 8K random to track pretty much the same as 4K random here, just at a lower rate because it’s moving twice the amount of data. At queue depths of sixteen or higher, our test subject delivers the most we’ve seen from a flash-based PCIe Gen4 SSD.

Supremely low latency at low queue depths here is exactly what we want to see from any SSD, and again, our D7-P5810 test subject delivers the best to date for any flash-based SSD we’ve ever tested. Outstanding.

128K Sequential Write/Read

We precondition the drive for 6,500 seconds, receiving performance data every second. Steady state for this test kicks in at 0 seconds. The average steady-state sequential write performance at QD256 is approximately 4,300 MB/s. We note our test subject is delivering exceptional consistency.

Again, and as we’ve seen every step of the way, our test subject delivers significantly more than its factory specs indicate.

6,800 MB/s is 400 MB/s more than factory spec and is impressive on its own. However, what we find most impressive is our test subject delivering more at queue depths of 1-4 than any flash-based PCIe Gen4 SSD we’ve tested to date. Additionally, at QD1, it’s the second-best we’ve ever recorded. This is supremely efficient and indicates to us that CSAL persistent write buffer implementation will be highly effective.

Benchmarks – Workloads

4K 7030

4K 7030 random is a commonly quoted workload performance metric for Enterprise SSDs.

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. Steady state is achieved at 2,000 seconds of preconditioning. The average steady-state performance at QD256 is approximately 700K IOPS.

Here, we find our test subject delivering well above its quoted up-to-spec of 645K IOPS at QD128. The performance curve here is stunning, with our test subject delivering the best we’ve ever seen from a flash-based SSD at queue depths of up to 16. The most performance where it matters most, you’ve got to love that.

Email Server

Our Email Server workload is a demanding 8K test with a 50 percent R/W distribution. This application gives a good indication of how well a drive will perform in a write-heavy workload environment.

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. Steady state is achieved at 2,000 seconds of preconditioning. The average steady-state performance at QD256 is approximately 310K IOPS.

This is a recurring theme. The D7-P5810 delivers more low queue depth performance than any other flash-based SSD we’ve ever tested, including the two PCIe Gen5 offerings that appear on this chart. Impressive.

OLTP/Database Server

Our On-Line Transaction Processing (OLTP) / Database workload is a demanding 8K test with a 66/33 percent R/W distribution. OLTP is online processing of financial transactions and high-frequency trading.

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. Steady state is achieved at 2,000 seconds of preconditioning. The average steady-state performance at QD256 is approximately 334K IOPS.

Everything stated about the previous test applies exactly here as well. The D7-P5810 is, at low queue depths, overall, the best-performing flash-based SSD we’ve ever tested. This is performance that matters.

Web Server

Our Web Server workload is a pure random read test with a wide range of file sizes, ranging from 512B to 512KB at varying percentage rates per file size.

We precondition the drive for 16,000 seconds, receiving performance data every second. We plot this data to observe the test subject’s descent into steady-state and to verify steady-state is in effect as we seamlessly transition into testing at queue depth. We precondition for this test with an inverted (all-write) workload, so no relevant information can be gleaned from this preconditioning other than verification of steady state.

Except for QD1 performance, nothing stands out as exceptional here. This is just fine by us because the D7-P5810 will never be utilized for this role, where ultra-high capacity is prized above all.

Final Thoughts

Caching QLC is what Solidigm created the D7-P5810 for. The drive has all the attributes necessary to, as we see it, literally, eliminate any of the typical drawbacks associated with QLC mass storage arrays. Throw in CSAL large write sequentialization/alignment persistence, and even QLC endurance deficiency is literally erased. This is the cherry on top.

Overall, the Solidigm D7-P5810 is the best-performing flash-based SSD we’ve ever tested at low queue depths, where performance matters most. Editor’s Choice.