Dataset · one three-node cluster per interface · testing ended 21 May 2021

At 128K, one three-node vSAN cluster read 10,366 MB/s on NVMe, 4,566 MB/s on SAS, and 2,776 MB/s on SATA

At 128K and 100 percent sequential read, one three-node VMware vSAN cluster reached 10,366 MB/s on KIOXIA CD6 NVMe, 4,566 MB/s on 12 Gbps SAS, and 2,776 MB/s on 6 Gbps SATA. Principled Technologies reported those cluster totals at each configuration's own 100 percent Vdbench IOrate. Hands-on testing ended 21 May 2021, the configurations were finalized on 14 April 2021 or earlier, and the KIOXIA-commissioned report is dated September 2021. The same rows show latency of 13.88 ms on NVMe and 3.51 ms on SATA, so the higher throughput is not a lower-latency result. The 6 Gbps and 12 Gbps figures are the printed link rates and are not converted to MB/s.

Download CSV11 rows · 17 columns · CSV

File /data/datasets/sata-sas-nvme-measured-throughput.csv · JSON metadata · All datasets · Storage throughput · SATA · SAS · NVMe · Theoretical link ceilings

Method

Population: one VMware vSAN cluster of three HPE ProLiant DL385 Gen10 Plus servers per interface, not a set of clusters. Science PDF Table 4 gives each node one disk group, two 3.84 TB capacity SSDs (six in the cluster), and one 800 GB KIOXIA KCM6XVUL800G NVMe cache. The cache model is the same on the SATA, SAS, and NVMe configurations. Capacity models are HP VK003840GWSRV (6 Gbps SATA), HPE VO003840RWUFB (12 Gbps SAS, the RM5 configuration), and KIOXIA KCD6XLUL3T84 (NVMe PCIe Gen 4). The hypervisor in the science PDF is VMware ESXi 7.0 Update 2, build 17630552. vSAN traffic used a dedicated 100Gb link with MTU 9,000.

Principled Technologies concluded hands-on testing on 21 May 2021. The printed results reflect configurations finalized on 14 April 2021 or earlier. The report is dated September 2021 and says the project was commissioned by KIOXIA. HCIBench 2.5.3 ran Vdbench 5.04.07. The parameter file sets rdpct=100, seekpct=0, xfersize 32k, 64k, and 128k, openflags=o_direct, warmup=30, elapsed=120, and curve=(10-120,10). The science PDF says the comparisons use the metrics at 100 percent I/O rate, "the point where the benchmark determined an optimal balance between higher I/O performance without incurring extreme storage latencies." The nine cluster rows are those printed Table 1 cells. Report Figures 1 through 3 print the same MB/s and call each figure the max throughput of a three-node cluster. This file does not read values off the bar drawings.

Link rate and measured throughput are separate columns. Table 4 prints 6 Gbps for the SATA capacity drive and 12 Gbps for the SAS capacity drive. The report says SATA "is limited to only a 6Gb/s connection" and that a 2021 SCSI Trade Association roadmap names both 12Gb/s and 24G SAS. The tested SAS link is the 12 Gbps label, not 24G. Neither rate is converted to MB/s on this page. A separate file,payload ceilings by generation, holds theoretical ceilings. Those ceilings are not copied into these cells.

The cluster MB/s and IOPS cells agree with a 1,048,576-byte megabyte and a 1,024-byte K, as arithmetic only. For 128K NVMe, 10366 × 8 = 82,928, the same integer as the IOPS cell. Across the nine cluster rows the largest gap is 11 IOPS (the 32K SATA row). The science PDF never defines MB. KIOXIA's FAQ does, for its own ratings: 1 MB is 1,000,000 bytes. The two MB labels are not treated as the same unit here.

What the MB/s column misses

The signal is cluster throughput at 100 percent IOrate. At 128K that signal is 10,366 MB/s on NVMe, 4,566 MB/s on SAS, and 2,776 MB/s on SATA. Dividing the printed cells gives 3.73 times, which the report rounds to "Up to 3.7X" and also calls more than three times the SATA configuration. The SAS cell is 64 percent above SATA, the increase the report prints.

The fault that signal misses is latency on the same Table 1 rows. At 32K the latencies are 1.05 ms, 1.19 ms, and 1.18 ms while throughput still ranks NVMe, then SAS, then SATA (5,269, 4,319, 2,693 MB/s). At 64K, SAS latency is 3.78 ms against SATA 2.16 ms and NVMe 1.42 ms, with throughput 4,562, 2,832, and 7,688 MB/s. At 128K, NVMe latency is 13.88 ms and SAS is 13.27 ms, against SATA 3.51 ms. The highest throughput is not the lowest latency. The science PDF does not publish the other points on the 10-to-120 percent curve, so these rows are not a common-latency comparison.

A second miss is the shape of the published parameter file. It lists ten storage definitions, but sd7 and sd9 both use lun=/dev/sdg, and sd8 and sd10 both use lun=/dev/sdh. The prose says 24 VMs with 10 data disks each, sized 10 GB. The file can prove those two device names are repeated. It cannot prove whether HCIBench rewrote the LUN list before the run, and it cannot prove that ten distinct disks were opened.

A third miss is the vendor rating beside the cluster total. KIOXIA's FAQ Table 3 rates one CD6-R drive at 3,840 GB as 6,200 MB/s sequential read at 128 KB and queue depth 32. That is a vendor claim. The FAQ does not print KCD6XLUL3T84, which is the capacity part in Table 4, and 10,366 MB/s is a six-drive cluster total under Vdbench's unlabeled MB. This table does not divide one by the other.

Limits

Population: one three-node VMware vSAN cluster per interface, not a sample of clusters. Each node is an HPE ProLiant DL385 Gen10 Plus with one disk group, two 3.84 TB capacity SSDs, and the same 800 GB KIOXIA KCM6XVUL800G NVMe cache, including the SATA configuration. Hands-on testing ended 21 May 2021. Configurations were finalized on 14 April 2021 or earlier. The report is dated September 2021 and was commissioned by KIOXIA. There is no confidence interval.
Not measured: random I/O, writes, mixed read/write, block sizes other than 32K, 64K, and 128K, 24G SAS, per-drive throughput, and a comparison held to one shared latency. The MB/s cells are each configuration's own 100 percent point on a 10-to-120 percent Vdbench IOrate curve. The other curve points are not in this table.
The report says vSAN services write I/O from the cache tier and read I/O from the capacity tier. It does not print a cache-hit counter. vSAN failures-to-tolerate and the RAID policy are not stated. The published parameter file repeats /dev/sdg on sd7 and sd9 and /dev/sdh on sd8 and sd10, while the prose says 24 VMs with ten 10 GB data disks, so the file does not show ten distinct LUNs.
Link rate and measured throughput stay in the units the sources print. 6 Gbps and 12 Gbps are not converted to MB/s. The PCIe 4.0 row is KIOXIA's April 2021 reprint of PCI-SIG rates. Science PDF Table 4 does not state the tested NVMe drive's lane count.
The 6,200 MB/s row is a KIOXIA vendor claim for one CD6-R drive at 3,840 GB, 128 KB, queue depth 32, with 1 MB defined as 1,000,000 bytes. The FAQ does not print part number KCD6XLUL3T84. The science PDF does not define the Vdbench MB, so the cluster MB/s cells are not divided by 6,200 here.
Science PDF Table 2 prints hardware cost and IOPS per dollar. Those figures are not rows in this file.

Table

11 rows, the same rows as the CSV. Units are in the column headers. A dash means that column is a different figure kind, not a missing measurement. Cluster MB/s are Principled Technologies' printed Vdbench units. The vendor MB/s uses KIOXIA's 1,000,000-byte megabyte. The Gbps column is the printed link rate and is not converted. Sources were opened on 2026-10-11.
Figure	Block size (as printed)	Interface	Cluster throughput (MB/s)	Cluster IOPS	Latency (ms)	Link rate (Gbps)	Vendor sequential read (MB/s)	PCIe transfer rate (GT/s)	PCIe x1 payload (GB/s)	PCIe x4 payload (GB/s)	Drive (as printed)	Link (as printed)	Population	Result date	Note	Source
measured_cluster	32K	SATA	2,693	86,165	1.18	6	—	—	—	—	HP VK003840GWSRV	6 Gbps, SATA, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 2 calls this MB/s the max throughput of a three-node vSAN cluster. The 6 Gbps cell is Table 4's SATA link label and is not converted to MB/s.	pt-science-0921
measured_cluster	32K	SAS	4,319	138,215	1.19	12	—	—	—	—	HPE VO003840RWUFB	12 Gbps, SAS, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 2 calls this MB/s the max throughput of a three-node vSAN cluster. The 12 Gbps cell is Table 4's SAS link label. The report names 24G only as a SCSI Trade Association roadmap rate, not this drive.	pt-science-0921
measured_cluster	32K	NVMe	5,269	168,609	1.05	—	—	—	—	—	KIOXIA KCD6XLUL3T84	NVMe PCIe Gen 4, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 2 calls this MB/s the max throughput of a three-node vSAN cluster. Table 4 prints NVMe PCIe Gen 4 and does not print GT/s or GB/s for this drive.	pt-science-0921
measured_cluster	64K	SATA	2,832	45,314	2.16	6	—	—	—	—	HP VK003840GWSRV	6 Gbps, SATA, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 3 calls this MB/s the max throughput of a three-node vSAN cluster. The 6 Gbps cell is Table 4's SATA link label and is not converted to MB/s.	pt-science-0921
measured_cluster	64K	SAS	4,562	72,994	3.78	12	—	—	—	—	HPE VO003840RWUFB	12 Gbps, SAS, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 3 calls this MB/s the max throughput of a three-node vSAN cluster. The 12 Gbps cell is Table 4's SAS link label, not the 24G roadmap rate.	pt-science-0921
measured_cluster	64K	NVMe	7,688	123,015	1.42	—	—	—	—	—	KIOXIA KCD6XLUL3T84	NVMe PCIe Gen 4, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. Report Figure 3 calls this MB/s the max throughput of a three-node vSAN cluster. Table 4 does not print a GT/s or GB/s rate for this PCIe Gen 4 drive.	pt-science-0921
measured_cluster	128K	SATA	2,776	22,211	3.51	6	—	—	—	—	HP VK003840GWSRV	6 Gbps, SATA, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. The report says the SATA configuration supported 2,776 MB/s on average across the workload window. Latency on this row is 3.51 ms. Figure 1 calls the MB/s a three-node cluster total.	pt-science-0921
measured_cluster	128K	SAS	4,566	36,526	13.27	12	—	—	—	—	HPE VO003840RWUFB	12 Gbps, SAS, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. The report says the value SAS configuration achieved an average of 4,566 MB/s, a 64 percent increase over SATA. Latency on this row is 13.27 ms.	pt-science-0921
measured_cluster	128K	NVMe	10,366	82,928	13.88	—	—	—	—	—	KIOXIA KCD6XLUL3T84	NVMe PCIe Gen 4, SSD, 2.5	One three-node VMware vSAN cluster; two 3.84 TB capacity SSDs per node (six); one 800 GB NVMe cache SSD per node; 100 percent sequential read	2021-05-21	Science PDF Table 1 at 100 percent Vdbench IOrate. The report says the NVMe configuration supported 10,366 MB/s, more than three times the SATA configuration, and prints Up to 3.7X. Latency on this row is 13.88 ms, higher than SATA. Cluster total, not a per-drive peak.	pt-science-0921
vendor_drive_rating	128 KB	NVMe	—	—	—	—	6,200	—	—	—	KIOXIA CD6-R Series 3840 GB	PCIe 4.0, NVMe 1.4	Single CD6-R drive vendor rating; 128 KB; queue depth 32	2021-04	Vendor claim from KIOXIA FAQ Table 3 for CD6-R at 3,840 GB: sequential read 6,200 MB/s at 128 KB and QD=32. Note 3 defines 1 MB as 1,000,000 bytes. Not a vSAN result. The FAQ does not print part number KCD6XLUL3T84.	kioxia-faq-2021
interface_line_rate	—	PCIe	—	—	—	—	—	16.0	1.969	7.88	—	PCIe 4.0	PCIe 4.0 interface rate reprinted by KIOXIA; not the vSAN cluster	2021-04	Not a drive measurement. KIOXIA FAQ Table 1, attributed to PCI-SIG: PCIe 4.0 at 16.0 GT/s, 1.969 GB/s on x1, and 7.88 GB/s on x4. The prose says approximately 2 GB/s per lane. Table 4 does not state the tested server's lane count.	kioxia-faq-2021

Columns

figure_kind (string): measured_cluster, vendor_drive_rating, or interface_line_rate. These are different populations and are not interchangeable.
block_size_as_printed (string): Block size as the source prints it: 32K, 64K, 128K, or 128 KB. Not converted to bytes.
interface (string): SATA, SAS, NVMe, or PCIe.
cluster_throughput_mb_s (number): Three-node vSAN cluster throughput in MB/s as Principled Technologies printed it at 100 percent Vdbench IOrate. Blank when the row is not that measurement. The PDF does not define MB.
cluster_iops (number): Cluster input/output operations per second at the same 100 percent IOrate point. Blank when the row is not that measurement.
latency_ms (number): Latency in milliseconds at the same 100 percent IOrate point. Blank when the row is not that measurement.
link_rate_gbps (number): Link rate in Gbps as science PDF Table 4 prints it: 6 for SATA or 12 for SAS. Not converted to MB/s. Blank when the source does not print a Gbps figure.
vendor_sequential_read_mb_s (number): Single-drive sequential read in MB/s. KIOXIA defines 1 MB as 1,000,000 bytes. A vendor claim, not the vSAN cluster. Blank on cluster rows.
pcie_transfer_rate_gt_s (number): PCIe transfer rate in gigatransfers per second from KIOXIA FAQ Table 1. An interface rate, not a drive measurement.
pcie_x1_payload_gb_s (number): PCIe x1 payload throughput in GB/s from KIOXIA FAQ Table 1. Not a Vdbench result.
pcie_x4_payload_gb_s (number): PCIe x4 payload throughput in GB/s from KIOXIA FAQ Table 1. Not evidence that the tested server used four lanes.
drive_as_printed (string): Capacity-drive model from science PDF Table 4, or the FAQ series name on the vendor row. Blank on the interface-rate row.
link_as_printed (string): Interface label copied from the source, including words the numeric columns do not encode.
population (string): What the row counts: one three-node cluster, one vendor drive rating, or a reprinted interface rate.
result_date (string): 2021-05-21 is the day Principled Technologies concluded hands-on testing. 2021-04 is the KIOXIA FAQ issue month, not a test day.
note (string): What the row is not, tied to a sentence in the cited file.
citation_id (string): Id of the opened source in the citations list.

License

Small derived table of figures printed in Principled Technologies' September 2021 Vdbench report and science PDF, and in KIOXIA's April 2021 PCIe 4.0 NVMe FAQ (Rev. 2.5). The infographic states Copyright 2021 Principled Technologies, Inc. The science PDF says the project was commissioned by KIOXIA and disclaims warranties on accuracy, completeness, and quality. Neither disclaimer grants an open-data license. KIOXIA copyright 2021 applies to the FAQ ratings. This file cites those cells and does not mirror the PDFs. Science PDF · Report PDF · KIOXIA FAQ PDF · Infographic PDF.

Sources

The science behind the report: Process more read-heavy I/O with HPE ProLiant DL385 Gen10 Plus servers configured with KIOXIA value SAS and NVMe mainstream SSDs (Principled Technologies, September 2021) (accessed 2026-10-11)
Process more read-heavy I/O with HPE ProLiant DL385 Gen10 Plus servers configured with KIOXIA value SAS and NVMe mainstream SSDs (Principled Technologies, September 2021) (accessed 2026-10-11)
What You Need To Know About PCIe 4.0 NVMe SSDs, KIOXIA FAQ, April 2021, Rev. 2.5 (accessed 2026-10-11)
Principled Technologies Vdbench infographic, September 2021 (accessed 2026-10-11)