Toshiba MG04 Enterprise 5TB HDD Review

1. Introduction
2. MG04 Enterprise 5TB HDD Internals and Specifications
3. Test System and Methodology
4. Benchmarks - 4k Random Read/Write
5. Benchmarks - 8k Random Read/Write
6. Benchmarks - 128k Sequential Read/Write
7. Benchmarks - Database/OLTP
8. Benchmarks - Email Server
9. Final Thoughts

Introduction

53

The MG04 series offers a capacity increase to 5TB for Toshiba's nearline datacenter HDD offerings. The new 5TB model features five 1TB platters that deliver a 24% increase in the data transfer rate (up to 205MB/s) in comparison to the previous generation MG03SCA400.

The 7,200-RPM MG04 series is available in 6Gb/s SAS and SATA models (MGO4SCA and MG04ACA, respectively), and is designed to satisfy workload requirements for cloud applications, mid-tier servers, and high-capacity data center storage systems. Other manufacturers are already offering 6TB products, and Toshiba has a second-generation MG04 coming in Q1 2015 that will bring them up to speed with their own 6TB product. The MG04 refresh will also add a 12Gb/s connection.

53

The MG04SCA products provide an UBER rating of one per 10^15, and feature a 64MB FIFO ring buffer. The series also features Persistent Write Cache technology (PWC). PWC leverages the EMF energy in the spindle motor (produced by spindle rotation) to power data-loss protection circuitry. This comes in particularly handy for Advanced Format 512e models that have read-write-modify data in cache during an unexpected power loss. PWC flushes read-write-modify data down to a non-volatile cache, and restores the data when power is re-applied to the drive.

The MG04 series offers 512e and 4K Native (4Kn) models, and our sample marks the first HDD with the 4Kn format in our lab. 4Kn HDDs first debuted in April of 2014, and are already spreading into the datacenter. The 4Kn format is required to address several challenges that have cropped up as HDD density has increased. 4k sector size increases format efficiency, and thus density, by requiring less space for redundant header areas. 512-byte sectors have also become physically smaller as drive capacity increases, which causes media defects of the same size to incur damage to a larger amount of data. This increases ECC requirements. 4K sectors ease error correction requirements for media defects, and also expand error correction capability by providing more room for ECC data.

The industry has already largely moved to physical 4K sectors with Advanced Format 512e drives. These drives expose themselves to the host as logical 512-byte sectors, but an internal emulation layer translates the data to the physical 4K sectors. This emulation layer can introduce a read-write-modify process when writing data, which negatively affects performance.

The 4Kn format exposes the sectors to the host logically, and physically, at its actual 4K size. This removes the translation layer, and provides performance benefits for files 4K and larger, but 512e drives still hold an advantage for smaller file sizes. Not all applications and operating systems have 100% compatibility with 4Kn drives as of yet, but compatibility is expanding as the industry bows to the inevitability of the 4Kn format. We caution readers to assure their OS, applications, and hardware are 4Kn compatible before deploying 4Kn HDDs.

The nature of the 4Kn format will provide performance advantages over the 512e Seagate and HGST HDDs in our test pool. Format and capacity variations require consideration during performance and power consumption analyses. Let's get the MG04SCA500A on the bench for our first look at a 4Kn HDD.

MG04 Enterprise 5TB HDD Internals and Specifications

MG04 Enterprise 5TB HDD Internals

The MG04SCA500A comes in the standard 3.5" form factor with normal fastener emplacements on the rear and sides. Other large capacity drives have transitioned to new fastener alignments that are not compatible with all drive trays.

53

A foam pad is placed between the PCB and the drive body to absorb vibration.

53

An LSI TT7050 controller rests in the top center of the PCB. There are accelerometers placed on the bottom left and middle of the right edge of the PCB to detect, and thus counteract, vibration in multi-drive enclosures.

53

The MG04SCA500A features a 6Gb/s SAS connection, with a step up to the 12Gb/s SAS connection slated for the MG04 refresh.

53

Idle power consumption is a pain point in the datacenter. Each manufacturer features proprietary implementations of enhanced idle modes that place the drive into deeper quasi-sleep cycles to conserve power. These modes can be enabled with a SAS Mode Page or SATA Set Feature command. Once enabled, these low-power idle modes place the drive into successively deeper idle states, but each consecutive sleep mode requires more time for resumption. We did not test with enhanced idle states enabled.

A key consideration for idle power consumption lies in the Watts-per-TB metric. These raw values represent the idle power consumption of each drive with no adjustments for capacity. Normalizing these values will bring the Seagate Enterprise Capacity into closer competitive range of the Toshiba MG04. The MG04 features an excellent idle power consumption of 8.2 Watts. The He6 provides the lowest idle power draw of the high-capacity segment.

MG04 Enterprise 5TB HDD Specifications

The 6Gb/s SAS Toshiba MG04SCA500A we are testing today features the 4Kn format, and there are 512E models available with the MG04SCAxxxE model numbers.

53

Test System and Methodology

53

53

Our approach to storage testing targets 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).

While under load, all storage solutions deliver variable levels of performance. 'Average' results do little to indicate 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. 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. This testing methodology illustrates performance variability, and includes average measurements during the measurement window.

IOPS data that ignores latency is useless. 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 utilize high-granularity I/O latency charts to illuminate performance during our test runs, and our Latency v IOPS testing reveals performance at varying latency thresholds.

We conduct our tests over the full LBA range to allow each HDD to highlight its average performance. All three HDDs spin at 7,200 RPM, but feature varying capacity points. The Toshiba MG04SCA500A also utilizes the 4K native format, while the HGST He6 and Seagate Enterprise Capacity utilize the 512e format. Format and capacity variations require consideration during performance and power consumption analyses. The first page of results will provide the 'key' to understanding and interpreting our test methodology.

Benchmarks - 4k Random Read/Write

4k Random Read/Write

53

Each level tested includes 300 data points (five minutes of one second reports) to illustrate performance variability. The line for each queue depth represents the average speed reported during the five-minute interval. We include a smaller embedded chart that lists performance at the highest load for easy interpretation.

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. 4K random performance is a heavily marketed figure, and is one of the most sought-after performance specifications.

The Toshiba MG04 takes a slight lead over the He6 in the random read workload. The MG04 also scales well by quickly delivering higher performance as we add load, and topping out at QD64.

53

53

Our Latency vs IOPS charts compare the amount of performance attained from each solution at specific latency measurements. Many applications have specific latency requirements. These charts present relevant metrics in an easy-to-read manner for readers who are familiar with their application requirements. The HDDs that are lowest and furthest to the right exhibit the most desirable latency characteristics.

The Toshiba MG04 delivers the most performance within a lower latency envelope.

53

The MG04 takes a commanding lead of 53 IOPS over the Seagate Enterprise Capacity. The He6, optimized for low power consumption, places a distant third with random write activity.

53

53

The MG04 continues to exhibit excellent Latency v IOPS performance.

53

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.

The MG04 leverages its impressive performance to deliver a win across the spectrum of mixed read/write random workloads. The He6 starts out with a higher random read speed than the Seagate Enterprise Capacity, but they trade positions as we mix in more write activity.

53

We record power consumption measurements during our test run at QD256. It is important to consider Watts-per-TB during power analysis due to the varying capacities in the test pool.

The MG04 draws less power than the Seagate Enterprise Capacity, but comes in a distant second to the He6.

53

53

We generate IOPS-to-Watts measurements from data recorded during the test period. The MG04 edges out the He6 in efficiency in the 4k write workload, which is quite the feat considering the He6's power-conservative slant.

Benchmarks - 8k Random Read/Write

8k Random Read/Write

53

Server workloads rely heavily upon 8k performance, and we include this as a standard with each evaluation. Many of our server workloads also test 8k performance with various mixed read/write workloads.

The MG04 exhibits a small lead over the He6 in this test.

53

53

The MG04 continues to provide excellent performance and latency in random read workloads.

53

The Seagate and HGST drives trade places with a pure random write workload, but the Toshiba MG04 remains on top during the test.

53

53

The MG04 continues to deliver enhanced latency performance.

53

The Toshiba MG04SCA500A takes the lead as we mix in more write activity.

53

The Toshiba MG04 and Seagate Enterprise Capacity are closely matched on power draw during the 4k write workload, while the HGST continues to lead with the lowest power consumption.

53

53

The MG04 again matches the He6's efficiency with random write workloads, but the He6 regains the lead with the random read workload.

Benchmarks - 128k Sequential Read/Write

128k Sequential Read/Write

53

We write sequentially to every LBA to highlight performance degradation from the outer to inner tracks of the drive. The drives begin with much higher speed on the outer tracks, and lose speed as they work inward. The varying capacity points in the test pool are evident, and the Toshiba MG04 requires the least time to complete a full pass across the platter due to its lower capacity. The Seagate and HGST models feature the same 6TB's of capacity, but the Seagate completes the test faster due to its increased sequential speed.

53

128k sequential speed reflects the maximum sequential throughput of the HDD, and is indicative of performance in OLAP, batch processing, streaming, content delivery applications, and backup scenarios. Today's HDDs are increasingly used for sequential workloads as SSDs encroach upon the application space.

The Toshiba MG04 and the Seagate Enterprise Capacity vie for the fastest speed in the 128k sequential testing, with the Seagate coming out on top. The He6 places a distant third.

53

53

The Seagate takes the lead in latency performance with sequential read activity.

53

Sequential write performance is important in tasks such as caching, replication, HPC, and database logging. The MG04 pulls off a slim lead of 1 MB/s over the Enterprise Capacity, and the He6 continues to lag behind with sequential workloads.

53

53

The Toshiba and the Seagate are closely matched, and the Seagate exhibits a distributed latency profile at higher queue depths. This variability disappears under sustained workloads.

53

The mixed-workload sequential test is one of our most demanding tests. Many drives will fare very well on the high-end, but fall short with mixed sequential activity. The Seagate wins six of the read-heavy categories, and the MG04 takes the lead in write-centric workloads.

53

The Toshiba MG04 provides incredibly low power consumption during the 128k workloads, so much so that we ran the tests three times to verify the results. The MG04 places second to the HGST He6, but draws much less power in comparison to the Seagate offering.

53

53

The impressively low power draw of the MG04, and its solid sequential performance, combine to provide a surprisingly competitive efficiency ratio compared to the power-optimized HGST He6.

Benchmarks - Database/OLTP

Database/OLTP

53

This test consists of Database and On-Line Transaction Processing (OLTP) workloads. OLTP is the processing of transactions such as credit cards, and high frequency trading in the financial sector. Databases are the bread and butter of many enterprise deployments. These demanding 8k random workloads with a 66 percent read and 33 percent write distribution bring even the best solutions down to earth.

The relatively large 8k file size lends itself well to the 4Kn format, and helps to deliver a big lead in the OLTP test.

53

53

The MG04 delivers an excellent Latency v IOPS ratio during the measurement window.

53

The Seagate Enterprise Capacity ekes out a win with lower power draw than the MG04, and the HGST continues to amaze with its extremely low power consumption.

53

Unsurprisingly, the He6 takes a commanding lead in the IOPS-to-Watts testing, but the MG04 also delivers great performance.

Benchmarks - Email Server

Email Server

53

The email server workload is a demanding 8K test with a 50% read and 50% write distribution. This application is indicative of performance in heavy random write workloads.

The MG04 delivers another solid win in the email server test with an average of 261 IOPS at QD256.

53

53

The big win in IOPS performance equates to a superb showing in the Latency v IOPS testing.

53

The Seagate and Toshiba vie for second place, with the MG04 taking a slight lead.

53

The MG04 continues to challenge the He6's efficiency advantage.

Final Thoughts

53

The Toshiba MG04 series provides robust performance metrics with a five-year warranty, and an innovative new power-loss protection scheme. Toshiba's PWC (Persistent Write Cache) leverages electromagnetic energy and non-volatile cache to protect data in transit during an unexpected power loss. This excellent addition provides an extra layer of protection for mission critical data.

Toshiba's move to 5TB signified a jump in capacity for Toshiba in comparison to their existing product lines, but other manufacturers have taken the lead, and are shipping 6TB drives. Toshiba is upgrading the connection to 12Gb/s SAS, and increasing density with their 6TB MG04 refresh in Q1 of 2015. The new MG04 will close the capacity gap with Toshiba's competitors to address the only shortcoming of the MG04 series.

The MG04 series delivered amazing performance in our testing, and routinely topped the charts in the majority of workloads. Our MG04 test sample features the 4Kn format, which provides a performance increase for larger files. The other drives in the test pool feature the 512e format, which provides an advantage with smaller file sizes. Some of the performance advantages observed in our testing can be chalked up to the 4Kn format, but many of the attractive features are simply the result of solid engineering.

The MG04 delivered leading speed in all of our 4k and 8k random tests, which fueled an optimal Latency-to-IOPS ratio that led the test pool. Mixed random workloads also benefited from the great performance. This translated well to our server workloads, and the MG04 took a comfortable lead in both tests.

Power consumption metrics revealed close competition with the Seagate Enterprise Capacity in random workload power draw. These comparisons are based upon the raw power consumption of the devices, but the Seagate and HGST HDDs offer an additional 1TB of capacity. This alters power-per-TB metrics, and requires consideration during power consumption analysis. The MG04 regularly rivaled the HGST He6 in random write workload IOPS-per-Watt, largely due to its very high performance in those workloads.

SSDs are altering the role of the HDD in the datacenter as they filter into caching and tiering roles. This shifts HDDs away from random workloads to sequential workloads. The MG04's impressive performance with sequential data tailors it well for a large variety of applications. The MG04 came close to unseating the He6 in sequential workload efficiency, which is surprising considering the HGST He6 is designed specifically to deliver low power consumption. The MG04 provided such surprising efficiency with sequential activity due to its incredibly low power draw during sequential access. The MG04 matched its low sequential power consumption with great overall sequential performance, which is quite the tandem.

Overall, the Toshiba MG04 provided superior performance and excellent power consumption metrics across the board. It is hard to place the MG04 for an award, specifically the Editor's Choice Award, largely because it is the first 4Kn drive in the lab. We need to do a deeper analysis of performance with other 4Kn drives in the test pool, so stay tuned for a future article with direct comparisons. We are also working on an article that explores performance variations between two of the same models with different formats.