sviko.com

Seagate Exos 10E2400 review - testing hybrid SFF HDD with advanced methodology

Many analysts believed that 2.5-inch hard drives in the future should have disappeared altogether, as in the corporate segment they would not find space between 3.5-inch drives and ultra-fast solid-state drives of different types (NVMe, SSD). But if you carefully examine the model range of the same SuperMicro, it is clear that more than half of the models of rack servers have 2.5-inch slots for HDD, and platforms for GPU computing generally come with only 2.5-inch compartments occupying less space. At this stage, analysts ’ forecasts did not come true - customers still have a demand for a large amount of local data on the server, which requires a large amount of storage.

Last year, Seagate rebranded its hard drives, combining and combining enterprise models into a series of Exos, including both 3.5-inch and 2.5-inch HDDs with different volume and different spindle speeds. The 2.4-terabyte 2.5-inch hard drive Exos E102400 (P/N ST2400MM0129) we are considering was previously called Enterprise Performance 10K v9, but the name change did not affect the technical characteristics or even the firmware, even the SKU number remained unchanged. As you can see from the previous name, we are dealing with the ninth generation of drives, of which the first six belonged to the Savvio 10K-10K series.6, seventh and eighth - to Enterprise Performance 10K, and ninth - to Enterprise Performance 10k and exos. Since the eighth generation, these drives have been added NAND-cache reading (TurboBoost Technology), so it is correct to call these drives hybrid.

Key features of Exos 10e2400 (Enterprise Performance 10k v9)

  • Volume - 2.4 TB
  • Interface - SAS-12
  • Native 4K sector support
  • Spindle speed - 10500 RPM
  • Typical access time is 2.9 MS
  • Buffer - 256 MB
  • NVC-cache - 8 MB
  • Flash read buffer - 16 GB
  • Size - 2.5 inches
  • 4 plates, 8 heads
  • Record type - PMR

Test bench configuration:

  • CPU - Intel Xeon E5-2603 v4
  • Motherboard: ASRock Rack EPC612D4U-2T8R
  • 32 GB of RAM
  • Debian 9 “stretch” (VDBench)
  • Seagate Savvio 10K.6 600Gb for comparison

Distinctive features of Exos 10E2400

One of the distinguishing features of the 9th generation 10K series is support for 512E (Advanced Format) and 4Kn formats. At the physical level, the sector size is 4 KB, but by default, the disk presents a 512-byte sector to the operating system through simple emulation. To date, not every operating system and not every program supports a 4-kilobyte logical sector. For example, even in VMware ESXi support for 4-kilobyte blocks appeared only in version 6.7, released in mid-2018, and still 4-kilobyte drives can not be forwarded to the guest OS via RDM. So, when you make sure that all your programs work properly with 4-kilobyte sectors, you will just need to reformat the hard drive to the new block size. Seagate promises that if the partitioning program supports Fast Format, the conversion will take seconds.

If you install the hard drive in the storage system, the storage system itself should be aware of the support for Fast Format, that is, these drives should be in the compatibility list. And of course, by default the disks come with a sector size of 512 KB in emulation mode (512E mode).

Efficiency of 8 Mb NVC Cache

Starting with the Seagate Enterprise Performance v8 series, hard drives have the protection of a small area of the write buffer against power loss. This is a small 8 MB non-volatile cache (NVC) consisting of eMLC memory cells, as in enterprise SSDS. If you connect a hard drive with Write Cache Enable = 0 (WCE=0), you still use write buffering, but the cache size is limited to the specified 8 megabytes. In case of a sudden power failure, the residual energy of the plates rotation is used, the motor switches to the generator mode and feeds the HDD electronics to write the buffer to the NVC area, where data can be stored for 90 days, and when the disk is turned on, the contents of the NVC cache are automatically transferred to magnetic media.

If the disk works with WCE=1, the buffer size is not limited to 8 megabytes, and you can expect better write performance, although not every test will show this difference. In fact, the NVC cache solves the age-old problem of no write buffering at the drive level in storage systems that almost always use a non-volatile memory SAS controller. Now we can say that write caching in exos 10E2400 disks is always used, just when WCE=1 buffer can exceed 8 MB, and when WCE=0 - no. Let’s see if it makes sense to talk about write buffering in real-world tasks at all.



Our tests show that write buffering gives any noticeable performance increase only in video surveillance systems, with sequential multithreaded recording, and in the old Savvio 10K drive.6 generally leads to a decrease in performance. I don’t think anyone in their right mind would use such an expensive disk for VMS software, so from our point of view, we can forget about NVC cache and built-in write buffering and not use it in further tests.

Turbo Boost Technology

Each disk Exos 7E2400 a 16-Gigabyte NAND cache on the eMLC chips are used to accelerate read-hot data. This technology looks quite simple and clear: the controller places the most frequently requested blocks in random read mode in the NAND-area, where data can be read with minimal delay compared to reading from magnetic plates. It is important to understand that TurboBoost Is not used for write and sequential read operations, but in RAID arrays the cache size will be summed, as each disk reads something different. Naturally, this technology is fully hardware, and no drivers or software for its operation is not necessary.

Here it is necessary to clearly understand that in the modern world of virtualization and Big Data, 16 GB is a drop in the sea, and modern storage and servers are perfectly able to cache and read and write - and SSD, and RAM. Of course, in arrays of 30 disks, the total volume of the TurboBoost area is already comparable to one cheap SSD of 480 GB, but still from a practical point of view - not impressive. However, think about how much data in the overall distribution is really “hot”? In typical databases - about 10%, or even less, so that the practical significance of TurboBoost entirely depends on the infrastructure topology and the nature of the requests.

Of course, if you use modern storage such as QNAP TDS-16489U, in which multilayer storage and SSD cache are offered for free, then TurboBoost from Seagate does not make sense. But if the storage does not support Tiering and SSD cache, or you install the hard drives in the machine under Windows Server or VMware ESXi, then of course TurboBoost can be the only opportunity to make frequently requested data on NAND memory, and it is completely transparent to the operating system, without additional licenses, programs and activation keys.

To test Turbo Boost, we limited the size of the test area to 16 GB, 32 GB, and 96 GB.



The results of random access are impressive. Here you have unattainable for conventional hard drives speed, but let’s increase the volume of the test area to 32 GB and see what happens.

Don’t forget during the break between the tests to drive the cache is a 3-hour random reading region of 500 GB.

The picture changes, and there are pronounced peaks and troughs.

Again, fill the cache with garbage and unfold the test in the 96-Gigabyte area.



As expected, with the growth of hot data, the effect of Turbo Boost comes to naught. There is an interesting point here: all enterprise Performance 10K v9 disks have the same NAND size, so hot data can be transferred to Exos 10E2400 of any size, taking into account the total amount of TurboBoost cache in the array.

Power consumption

Seagate series Exos 10E2400 have 4 gradation of power. Depending on the load, the drive may be in one of 4 States:

  • The heads above the disk, the controller reduced power consumption (4.9 W, the transition to a condition - 100 MS)
  • Heads are parked, disks rotate with the maximum speed (3.78 W)
  • The heads are parked, the discs rotate at a reduced speed (2.13 W)
  • Head is parked, the motor is stopped (1.24 W)

For read and write operations, the power consumption of the disk varies from 6.98 to 8.75 watts.

Reliability and temperature performance

Seagate Exos 10E2400 physical activity when ambient temperature is below 26 degrees Celsius. The claimed HDD resource is 2 million hours (228 years) and at least 600,000 head Parking operations. The documentation for the disk says that the reliability indicators are calculated for the temperature of the HDD below 50 degrees Celsius, that is, for Exos 10E2400, 49 degrees and below - this is the norm, and the limit heating, judging by the SMART, is 60 degrees Celsius.

For some of its hard drives, such as the Ironwolf Pro series, the manufacturer specifies the normal load in terabytes per year, including read and write operations. For the enterprise Performance HDD series, there is no such parameter, and it is especially interesting how the NAND part of the hard drive will age. Seagate strongly bypasses the issue of the resource of the chip used for the technology with sed, only mentioning that the eMLC memory works in such a way that its resource exceeds a multiple of the indicators of the MLC chips in a conventional SSD drives and are designed for the whole lifetime of the disk.

Whatever it was, the Exos 10E2400 series are provided with a 5 year warranty without any restrictions for the load.

Testing

As you have seen, the results of the hybrid hard drive speed will depend heavily on the size of the active region in which the reading takes place. In our basic tests, we use the entire disk capacity to show the overall performance of the Exos 7E2400. Let’s start with the traditional random access test.it was, on the EOS 10E2400 series hard drives provided a 5-year warranty without any restrictions on the load.


Let’s check how random access performance changes depending on the size of the transaction.



A huge drop in write performance with the growth of the transaction is higher 4Kb can be attributed to its own 4-KB sector size drive. This advantage is held on transactions up to 64 KB and then nullified. In operations of casual reading, in General, against physics you will not go: the bigger volume of a disk gives advantage only in area above 512 KB, that is in area of work of VMware ESXi which by default marks out the file system VMFS6 with the size of sector 1 MB.

Let’s look at sequential access.


Very good!

It is much better to see how sequential and random access are combined in real-world tasks.



Random write operations - the weak side of the considered hard drive, so all three parameters (speed, bandwidth, delay), he loses to his ancient colleague.

As I mentioned above, no one in their right mind would buy 2.4-terabyte hard drives for video surveillance, but in a large installation, a portion of the array can be dedicated to storing VMS data and initial analysis of records by AI. We saw that in the test Macroscop novelty gives about 88 megabytes per second, which corresponds to 100 streams from 1080p cameras. Let’s check the delay for different number of cameras.


As can be seen from the diagram, in the area of 100 cameras, Eos 10E2400 gives approximately two-fold superiority in access delay. In simple terms, where the old hard drives VS-software will begin to swear on the error of access to the disk system, the new hard drives will run smoothly.

Capacity density and issue price

In standard 2U cases, designed for 24 drives, Exos 7E2400 drives can achieve a density of 28.8 TB of raw capacity per 1 unit. In the same JBOD-case for 12 LFF disks, 3.5-inch Exos X12 with a volume of 12 TB will give three times the density - 72 TB per 1 unit. One hard drive Exos 7E2400 (p/n ST2400MM0129) has a cost of $460, and a 12-terabyte 3.5-inch Exos X12 ST12000NM0027 - $450, so that the direct comparison of the price difference per terabyte just blatant, but let’s see how to change the situation in the array, why take 2U shelf and fill it with hard drives.

For a 24-disk SFF shelf we will have the following configuration: RAID 6 on 22 hard drives + 2 hot swap drives. This will give us 48tb of raw capacity on magnetic media plus 352 Gb of cache in SSD area for a total cost of $11040.

In the 16-disk LFF shelf for the same volume, we need to install 8 exos X12 hard drives, of which 6 will be in RAID 6 and 2 in Hot Spare. To add a read NAND cache, we will need to install a SAS SSD, such as the Seagate Nytro ST400FM0233 with a capacity of 400 GB and a price of $743. The cost of this solution will be $4342.

In terms of volume, 2.5-inch 10K hybrid hard drives are 2.5 times smaller than 3.5-inch HDDs and SSDS.

Conclusions

In addition to the volume, which is steadily increasing every year, we offer new technologies that slowly penetrate into the hard drives. According to the test results of Exos 10E2400, we can draw the following conclusions: the disk shows excellent results for HDD random reading with a transaction size of 4K and random write with a transaction size of 1Mb. But the patterns of real-world problems do not confirm this assumption, and the only serious speed advantage over its distant ancestor Exos 10E2400 showed in the import of 4K video.

However, everything changes if the nature of the load changes. In real life, it is unlikely that you will keep a database of 2 TB on one hard drive, and as soon as the number of disks in the array increases, and the often requested data will be indexed and cached in the NAND-area, turbobost will come into play, which can give a speed increase of 1.5-2 times in mixed operations and 4-5 times in read operations, on any operating system and with any controller.

It is also worth noting that on 8 disks in RAID 10, you can achieve a speed of 1 GB/s, which will allow you to work with 4K video without real-time compression on a volume of 9.6 TB. In machines that work with AI applications for video analysis and in media studios, this is one of those cases where you can collect a relatively inexpensive array with a crazy linear speed.

There is another point that I can not ignore: in the enterprise Performance 10K v8 series, 0.9-1.8 TB disks had 32 GB of NAND cache, and in 10K V9 for some reason the cache was reduced to 16 GB, so when choosing models with a capacity of 900 to 1.8 TB under transactional load, it makes sense to look at the previous, eighth generation.