KingFast, based out of Shenzhen China, recently announced a merger (April 25, 2013) with the much larger SSD manufacturer RunCore. For now, KingFast is operating as a sub-brand of RunCore. In the future, we would expect there to be some consolidation of the product lines. RunCore is a much larger company, currently ranked as the largest SSD manufacturer in China and one of the largest in the world. RunCore has been growing at a tremendous rate with a revenue growth rate of 700% in 2012. The rapid expansion of RunCore and its high-end SSD products meshes well with the consumer and value-priced industrial SSD's that KingFast brings to the partnership.
The KingFast F3 Plus Series KF2510SCF SSD merges the LSI SandForce SF-2281 controller and Intel 25nm SLC NAND into one package. This arrangement features sequential read/write speeds of 556/530 MB/s, respectively. The random read IOPS are 55,735 and random write weighs in at a higher 83,221 IOPS. These are likely FOB (Fresh out of Box) numbers with 100% compressible data.
Recently we published a review of the SuperSSpeed S301 Hyper Gold Enterprise SSD, which also features a very similar build with the SF-2281 and Intel SLC flash. The real differentiator between these two products is that the SuperSSpeed is geared specifically for the client-side, and the KingFast F3 Plus Series SSD was designed and optimized from the ground up for datacenter applications.
With the same controller and NAND on both of these SSD's, we will put them head to head to see which can provide the best solution in an enterprise environment. There are always different variables to take into consideration beyond the normal 'feeds and speeds' when testing enterprise storage solutions. Typically, reliability and enhanced efficiency will trump less-refined solutions, even if they are faster.
The emergence of these value-oriented SLC SSD's brings forth an interesting product for those seeking tremendous endurance specifications at lower price points. The KingFast F3 Plus retails for roughly $2.50 per GB at the time of publishing for customers purchasing 10 or more SSD's. This is in the same price range of the SuperSSpeed offering, with its $2.00 per GB price. Part of the pricing disparity is due to the slightly larger capacity of the SuperSSpeed S301. The SuperSSpeed S301 features 128GB in comparison to the 120GB of the KingFast F3 Plus.
The reason for the capacity disparity is due to R.A.I.S.E functionality onboard the KingFast F3 Series SSD. The SuperSSpeed has this feature disabled. The SF-2281 controller is intended for consumer applications, but the use of R.A.I.S.E. functionality significantly boosts the capability of the SSD to deal with uncorrectable errors. This feature embeds a layer of parity in the NAND to allow for recovery from a failed sector, page, or even an entire block.
Enabling the R.A.I.S.E. feature creates a small loss of capacity, but yields huge gains in reliability. In conjunction with the 55b/512 BCH ECC this boosts the Uncorrectable Bit Error Rate (UBER) to 10-29, which equates to a quadrillion fewer uncorrectable errors than other controllers. Pairing this vigorous level of data protection with SLC NAND and its impressive 100,000 P/E cycles creates a very durable SSD that should handle the rigors of even the heaviest write workloads with ease.
Today in the comparisons of the two LSI SandForce SSD's we will learn that things aren't always what they seem when comparing raw speed of several SSD's. Even when performance results are mind-numbingly similar during the tests, there can be large differences in power and efficiency. With efficiency dictating the shape of today's datacenters, our results bring into focus the power efficiency metrics between two very similar SSD's.
KingFast F3 Plus Series Specifications
The KingFast F3 Plus Series SSD features an impressive endurance specification of 6,000 TBW for the 60GB model and 12,000 TBW for the 120GB model. There is no information on how these endurance numbers were obtained, but with SLC, it is certain that the F3 Plus will offer up plenty of endurance for users.
The KingFast F3 Plus features sequential read/write speeds of 556/530 MB/s, respectively. The random read IOPS are 55,735 and random write weighs in at a higher 83,221 IOPS. These are FOB (Fresh out of Box) numbers with 100% compressible data. We will test the steady state performance with 50% compressible data in the following pages.
Power consumption is rated at .3W idle and 2W under a heavy random 4K write workload. These seem to be very optimistic power consumption figures for a heavy write load, and we will put this to the test in this evaluation.
The KingFast F3 Plus sports 55b/512 BCH ECC and enabled R.A.I.S.E functionality to protect user data. There are no capacitors for data hold up with the F3 Plus, but for LSI SandForce powered SSD's power hold-up features aren't as important as with controllers that utilize DRAM caching.
Forgoing the use of a DRAM cache simplifies the construction of the SandForce SSD's, but also helps to eliminate much of the damage with unexpected power loss. LSI SandForce SSD's have a 1 in a 1,000 risk of losing data during a power loss event. While this does not entirely negate the need for supercaps, there is less risk in operating without power hold-up due to the inherent DRAM-less design of the LSI SandForce controllers. The addition of LSI SandForce compression algorithms reduces the wear on NAND during the life of the SSD. When combined with SLC flash this should create a superior SSD in terms of endurance and durability.
Surprisingly the KingFast F3 Plus only features a three year warranty. With many MLC SSD's from the current crop of contenders warrantied for five years, we feel that this SLC SSD should have a longer warranty period.
KingFast F3 Plus Series Internals
The KingFast F3 Plus features branding on the front of the SSD and the rear is devoid of any markings. The SSD's come in either 9mm or 7mm Z-Heights and the standard 2.5" form factor.
The F3 Plus comes in a sturdy molded alloy case. There are no thermal pads, but with the very high heat tolerance of SLC, this isn't required for the banks of NAND. The LSI SandForce controller has built-in thermal protection, which will throttle the SSD at a temperature threshold.
We also observe the four packages of 29F16B16NCNE1 Intel 25nm SLC NAND.
The NAND and controller populate the top of the PCB and we can also observe the robust power components onboard the PCB. There is no DRAM cache for LSI SandForce SSD's, which provide them an advantage in power loss events.
The bottom of the PCB is unremarkable, with all of the NAND on the top of the PCB.
Test System and Methodology
We utilize a new approach to HDD and SSD storage testing for our Enterprise Test Bench, designed specifically to target long-term performance with a high level of granularity.
Many testing methods record peak and average measurements during the test period. These average values give a basic understanding of performance, but fall short in providing the clearest view possible of I/O QoS (Quality of Service).
'Average' results do little to indicate the performance variability experienced during actual deployment. The degree of variability is especially pertinent, as many applications can hang or lag as they wait for I/O requests to complete. This testing methodology illustrates performance variability, and includes average measurements, during the measurement window.
While under load, all storage solutions deliver variable levels of performance. While this fluctuation is normal, the degree of variability is what separates enterprise storage solutions from typical client-side hardware. Providing ongoing measurements from our workloads with one-second reporting intervals illustrates product differentiation in relation to I/O QOS. Scatter charts give readers a basic understanding of I/O latency distribution without directly observing numerous graphs.
Consistent latency is the goal of every storage solution, and measurements such as Maximum Latency only illuminate the single longest I/O received during testing. This can be misleading, as a single 'outlying I/O' can skew the view of an otherwise superb solution. Standard Deviation measurements consider latency distribution, but do not always effectively illustrate I/O distribution with enough granularity to provide a clear picture of system performance. We use histograms to illuminate the latency of every single I/O issued during our test runs.
Our testing regimen follows SNIA principles to ensure consistent, repeatable testing. We attain steady state through a process that brings the device within a performance level that does not range more than 20% during the measurement window. Forcing the device to perform a read-write-modify procedure for new I/O triggers all garbage collection and housekeeping algorithms, highlighting the real performance of the solution.
We measure power consumption during precondition runs. This provides measurements in time-based fashion, with results every second, to illuminate the behavior of power consumption in steady state conditions. Power consumption can cost more over the life of the device than the initial acquisition price of the hardware itself. This significantly affects the TCO of the storage solution. We also present IOPS-to-Watts measurements to highlight the efficiency of the storage solution.
For LSI SandForce-powered SSD's we conduct all preconditioning, and the following workloads, with 50% compressible data. This is the result of industry feedback, and the middle ground provides a picture of performance that is neither too critical, nor too favorable. Readers must bear in mind that there is varying performance with LSI SandForce processors due to its internal compression engine. Incompressible data will score much lower than the results presented in these tests, and fully compressible data will score much higher. It is important for administrators to know their workload and select the storage solution accordingly.
Our test pool features three SSD's of varying capacity, so it is important to bear this in mind when viewing results. We are comparing two very similar LSI SandForce SSD's head-to-head, and adding in an MLC SSD as a comparison sample. The first page of results will provide the 'key' to understanding and interpreting our new test methodology.
4K Random Read/Write
We precondition the KingFast F3 Plus Series SSD with 50% compressible data for 18,000 seconds, or five hours, receiving reports on several parameters of workload performance every second. We then plot this data to illustrate the drives' descent into steady state. Please refer to the previous page for a note on data compressibility.
This chart consists of 36,000 data points. This is a dual-axis chart with the IOPS on the left and the latency on the right. The teal dots signify IOPS during the test, and the dark blue dots are latency measurements during the test period. We plot latency data in a logarithmic scale to bring it into comparison range. The lines through the data scatter are the average during the test. This type of testing presents standard deviation and maximum/minimum I/O in a visual manner.
Note that the IOPS and Latency figures are nearly mirror images of each other. This illustrates the point that high-granularity testing can give our readers a good feel for the latency distribution by viewing IOPS at one-second intervals. This should be in mind when viewing our test results.
We provide histograms for further latency granularity below. This downward slope of performance happens very few times in the lifetime of the device, typically during the first few hours of use, and we present the precondition results only to confirm steady state convergence.
Each QD for every parameter tested includes 300 data points (five minutes of one second reports) to illustrate the degree of performance variability. The line for each QD represents the average speed reported during the five-minute interval.
4K random speed measurements are an important metric when comparing drive performance, as the hardest type of file access for any storage solution to master is small-file random. One of the most sought-after performance specifications, 4K random performance is a heavily marketed figure.
The KingFast F3 Plus averages 55,348 IOPS with a 4K random workload at QD256. The SuperSSpeed S301 falls into such a close race with the F3 Plus that the results overlap each other. The Optimus, used as an MLC comparison, rises above 90,000 IOPS.
Garbage collection routines are more pronounced in heavy write workloads. This leads to more variability in performance. Both LSI SandForce powered SSD's exhibit variability during the 4K write testing. This is due to the compression engine working furiously to compress the data as much as possible in the background. We are using 50% compressible data in all of our tests, and more compressible workloads experience significantly less variability. The KingFast averages 51,390 IOPS during the QD 256 test measurement window.
Our write percentage testing illustrates the varying performance of each solution with mixed workloads. The 100% column to the right is a pure write workload of the 4K file size, and 0% represents a pure 4K read workload.
Both the F3 Plus and the S301 exhibit very similar performance, and great performance with a heavy write workload.
The F3 Plus and S3901 run neck and neck in the latency department, with the S301 delivering slightly more IOPS at the higher range. The F3 Plus provides 78% of I/O in the 4-6ms range, and 18% in the 6-8ms range.
We record the power consumption measurements during our precondition run. We calculate the stated average results during the last five minutes of the test, after the device has settled into steady state.
The real differentiators begin to show with our power consumption and efficiency tests. The F3 Plus delivers much lower wattage than the other SSD's, pulling an average of 3.5 Watts during the preconditioning period.
IOPS to Watts measurements are generated from data recorded during our precondition run, and the stated average is from the last five minutes of the test. The F3 Plus delivers an outstanding 15,889 IOPS per Watt in this test, significantly ahead of the S301 and Optimus.
8K Random Read/Write
8K random read and write speed is a metric that is not tested for consumer use, but for enterprise environments this is an important aspect of performance. With several different workloads relying heavily upon 8K performance, we include this as a standard with each evaluation. Many of our Server Emulations below will also test 8K performance with various mixed read/write workloads.
The average 8K random read speed of the KingFast F3 Plus at 256 QD is 38,040 IOPS, which again mirrors the SuperSSpeed S301 closely.
The average 8K random write speed of the F3 Plus averages 31,865 IOPS. The wide performance variability is indicative of LSI SandForce characteristics with partially compressible data.
The F3 Plus scores similarly to the S301 in this test, with the Optimus dipping below the two SLC SSD's.
The KingFast F3 Plus provides 68% of requests at 6-8ms, and 27% at 10-20ms.
Power consumption for the F3 Plus comes in at 3.03 Watts.
The KingFast F3 Plus averages 10,515 IOPS per Watt in this test, easily beating the other SSD's.
128K Sequential Read/Write
The 128K sequential speeds reflect the maximum sequential throughput of the SSD using a realistic file size encountered in an enterprise scenario.
The KingFast F3 Plus and the SuperSSpeed S301 run neck-and-neck in this test scenario. The results are so close that for the higher QD the results for the F3 Plus are behind the S301. The F3 Plus averages 531MB/s in this test.
The F3 Plus and the S301 exhibit very similar performance in this test, averaging 477MB/s.
The F3 Plus and S301 again run neck and neck with virtually indistinguishable performance results.
The F3 Plus provides 99.8% of operations (11,401,122 I/O's) in the 60-80ms range, with the remaining 1,964 IOPS occurring in the 80-100ms range.
The F3 Plus averages a miserly 2.6 Watts during the measurement window.
The F3 Plus easily tops the competition in the MB/s per Watt test, averaging 182MB/s per Watt.
Database/OLTP and Webserver
This test emulates Database and On-Line Transaction Processing (OLTP) workloads. OLTP is in essence the processing of transactions such as credit cards and high frequency trading in the financial sector. Enterprise SSD's are uniquely well suited for the financial sector with their low latency and high random workload performance. Databases are the bread and butter of many enterprise deployments. These are demanding 8K random workloads with a 66% read and 33% write distribution that can bring even the highest performing solutions down to earth.
As we begin to delve into our mixed workload benchmarks it is important to note, again, that we test the KingFast F3 Plus with 50% compressible data. OLTP and Database workloads tend to be highly compressible, so in real deployment performance may be higher than represented in these tests. Other workloads we test may also tend to have different compression ratios, so it is important to understand the workload and adjust purchasing decisions accordingly.
The KingFast F3 Plus averages 35,584 IOPS at QD 256, slightly edging out the SuperSSpeed S301, but falling to the Optimus.
The F3 Plus delivers 5,553,504 I/O's (52.2%) at 6-8ms, 2,370,488 I/O's (22.2%) at 4-6ms, and 2,288,988 I/O's (21.5%) at 8-10ms.
The F3 Plus again exhibits excellent power consumption metrics, averaging 2.8 Watts during the measurement window.
The KingFast F3 Plus is just as efficient an SSD as they come, averaging 12,839 IOPS per Watt.
The Webserver profile is a read-only test with a wide range of file sizes. Web servers are responsible for generating content for users to view over the internet, much like the very page you are reading. The speed of the underlying storage system has a massive impact on the speed and responsiveness of the server that is hosting the website, and thus the end-user experience.
The KingFast F3 Plus falls into a similar performance range with the S301, providing an average of 22,074 IOPS at QD 256.
The F3 Plus provides 5,154,601 I/Os (78%) in the 10-20ms range, and 1,346,964 I/Os (20.4%) in the 8-10ms range.
The F3 Plus averages 2.75 Watts during the measurement window.
The F3 Plus averages 5,430 IOPS per Watt.
Fileserver and Emailserver
The File Server profile represents typical file server workloads. This profile tests a wide variety of different file sizes simultaneously, with an 80% read and 20% write distribution.
The F3 Plus averages 29,427 IOPS at QD 256.
The F3 Plus provides 6,650,588 I/Os (74%) in the 8-10ms range, and 1,905,531 I/Os (21%) in the 8-10ms range.
The KingFast F3 Plus averages 2.63 Watts in steady state.
The F3 Plus averages 11,283 IOPS per Watt, beating the other two contenders.
The Emailserver profile is a very demanding 8K test with a 50% read and 50% write distribution. This application is indicative of the performance of the solution in heavy write workloads.
The F3 Plus averages 33,691 IOPS at QD 256.
The F3 Plus provides 53.2% of I/O's in the 6-8ms range, 30.1% at 8-10ms, and 11.5% at 10-20ms.
The F3 Plus averages 2.96 Watts during the measurement window.
The KingFast F3 Plus closes out our testing with an average of 11,439 IOPS per Watt.
Perhaps one of the largest motivating factors for administrators looking to deploy SSD's into datacenter applications is the huge savings in power, space and cooling. The KingFast F3 Series SSD sips power at a much lower rate than many of the top-notch SSD's available today. With its excellent speed and low power consumption, the F3 Plus delivers a great IOPS-to-Watt ratio across the board.
There was a tremendous difference in the power consumption of the KingFast F3 Plus Series SSD in comparison to the very similar SuperSSpeed S301 Hyper Gold SSD. For two SSD's that share so many similarities, with the exact same NAND and controller, it was interesting to see the huge difference in power consumption and efficiency.
The KingFast F3 Plus Series SSD performed admirably in our testing with the superb write performance expected of an SSD with SLC NAND. There are areas where the KingFast F3 Plus can improve though, and perhaps the most pressing need would be the three year warranty. Comparable MLC products offer warranties of five years, so the three year warranty is a bit on the low side for enterprise applications. With the extreme endurance and durability offered by the SLC NAND, it should be a relatively safe bet to guarantee the SSD for five years even in write intensive environments.
The other concern is capacity. The F3 Plus only comes in two capacities of 60GB and 120GB. This limits the number of applications that can benefit from such small storage devices. These SSD's will provide great performance in RAID arrays, due to their higher write endurance, which would allow for an aggregation of available capacity. Some will require more capacity and will be forced to look at other products.
One of the misguided views we hear frequently is that there is no differentiation from one LSI SandForce product to another. Our testing today shows that this is simply not true. While LSI SandForce provides turnkey firmware's and reference PCB designs, many manufacturers use different methods, components, and at times completely proprietary PCB's.
This provides the manufacturer with a flexible platform for deployment, but the quality of the components used and different PCB designs can affect the performance of the solution in other ways. As out testing today highlights, there is a very big difference between two very similar LSI SandForce SSD's in power consumption metrics.
Why is efficiency so important? Over the life of the device, it can cost more to power and cool the storage solution than the initial acquisition cost. When deployed en masse into datacenters, small differences in power from one drive can become huge differences when multiplied by hundreds or thousands. The largest ongoing expense in any size deployment is the cost of power, and when making a decision between two devices, power consumption is always a concern. Many enterprise product evaluations do not provide any insight into power and efficiency metrics, ignoring one of the most important facets of any storage device.
RunCore recently secured $10 million in Series B funding for investment into research and development programs. The merger with KingFast also brings more experience and engineers into the picture, and with the latest infusion of cash, we expect to see KingFast continue to perform well in the long term. The KingFast F3 Plus Series SSD proves to be an efficient SSD with great performance and at a lower price point.