Myce/OakGate 4K Read and Write Latency Tests
These tests steadily increase the random 4K IO demand in
terms of IOPS, and report the drive’s response in terms of Average IOPS, Average
Latency and Maximum Latency. We also use the detailed Latency results for each
IOPS level to extract Quality of Service data. It is designed to show a
drive’s maximum IOPS capability and report the all important Latency numbers
for each level of IOPS demanded. The Maximum latency numbers give us an
insight into the occurrence of Latency peaks that could cause an unexpected
response from time to time.
Firstly, here are the results for the initial
Pre-Conditioning step (4K Random Writes) –
Myce/OakGate 4K Read and Write Latency Tests
Please note that the 4K Random Writes were preceded by a
pre-fill step, which performed 128K sequential writes to twice the drive’s capacity
(to facilitate the achievement of a steady state). You can see that it takes a
long time for the Micron 5100 ECO 1920GB to reach a steady state. Even after 10
hours the Micron 5100 ECO is just hitting a Steady State of around 28,500 IOPS.
4K Latency Read Test
Myce/OakGate 4K Read Latency Test
We can see that the drive can no longer meet the increase in
IOPS demand beyond 75,000 IOPS.
We can see that read latency remains below 420 microseconds
all the way up to its maximum IOPS level. This is an excellent result.
We can see some Maximum Latency spikes at the 21,000, 47,000,
and 73,000 IOPS levels. We saw the same seemingly regular frequency of these
spikes with the 5100 MAX and it suggests that they may be caused by a regular
event in the drive’s housekeeping routines.
Let’s have a close look at the distribution of the Latency
results at the 21,000 IOPS level (one of the spikes) –
As this is the first time in this review, that we are
looking at a High Resolution Latency Histogram, here’s an explanation – The X
axis to the left is the count of the IOs in the observation period (in a Round)
that had a Latency of the value along the Y axis (please note that the X axis
is logarithmic to allow the low order counts of the huge number of IOs that
have been measured to be visible); the Y axis is the Latency value measured in
Microseconds; The X axis to the right is the % of the Total IOs observed that
have a Latency <= to a given Latency value; the rate of getting to 100% is
highlighted by the red graph line.
We can see that 99.9% of the Latency values are <= 200 Microseconds
and there are remarkably few outliers – the Quality of Service as measured in
this test is outstanding.
4K Latency Write Test
Myce/OakGate 4K Write Latency Test
We can see that in this test the drive continues to meet the
increase in demand up to a level of 32,000 IOPS.
Here we can see that Average Write Latency stays below 50 microseconds
all the way up to a demand of 31,000 IOPS – an outstanding result.
You can see here that there is one notable Maximum Latency
peak at the level of 30,000 IOPS.
So let’s have a look at the distribution of the Latency
Values at the 30,000 IOPS level –
We can see that 99.9% of the Latency Values are <= 1,220 Microseconds,
and there is only a relatively small number of outliers. This is an outstanding
Quality of Service
Quality of Service (QoS) of an SSD consists of the
predictability of low latency (and consistency of high IOPS) while servicing a
particular IO workload. This means the latency needs to be within a specified
range without having unexpected outliers.
QoS of an SSD is most often measured as the latency achieved
for a workload for a specified percentage of the time. An SSD that has a lower
latency for a percentage will exhibit a superior level of performance
predictability. QoS is most often measured (and specified by Manufacturers) for
4K Random Reads and Writes at a Queue Depth of 1 and/or 32.
Commonly the cause of an outlier is the background tasks performed
by an SSD’s controller, such as required to reclaim spare space and to perform
wear levelling, which may consume much of an SSD’s bandwidth temporarily
starving host IOs thus causing significant latency variation.
It is also important to note that QoS will vary depending
upon the level of IOs (IOPS) being serviced. Not surprisingly, as the IOPS
level increases the QoS generally drops (though there could be exceptions to
this, for example, if an SSD’s controller initiates background processes when
it is less busy the occurrence of outliers may actually be greater at lower
levels of demand)
Let’s have a closer look at the Quality of Service delivered
by the Micron 5100 ECO.
From the above test for 4K Random Reads I have extracted the
95th, 99th, 99.9th and 99.999th
percentile Latency values for levels of IOPS demand in the range of 5,000
through 75,000 IOPS, in increments of 5,000 IOPs, and charted them below. The
Latency Value (percentile rank) for the 95th percentile is the
Latency that 95% of all IOs fall within (and similarly for the 99th,
99.9th and 99.999th percentiles). The percentiles are a
measure of the drives consistency in the delivery of its performance. Here are
the results for Random 4K Reads –
The Quality of Service at the given IOPS levels can then be
compared to that achieved by other drives. For example, here are the values
for the outstanding Samsung PM863 960GB –
You can see that the Samsung PM863 is consistently better
than the Micron 5100 ECO. Having said this, it is fair to say that the Micron
5100 ECO delivers excellent results.
Similarly, for 4K Random Writes I have extracted the 95th,
99th, 99.9th and 99.999th percentile Latency
values for the range of 5,000 through 65,000 IOPS, in increments of 5,000 IOPs,
and charted them as follows –
These are outstanding results.
Now, let’s compare them to the results that we recorded for
the Samsung PM863 960GB –
The results for the Samsung PM863 are only charted to 15,000
IOPS as the drive topped out at 19,000 IOPS.
Up to 15,000 IOPS the Samsung is consistently better but it
is important to emphasise that the Micron 5100 ECO also has outstanding results
and reaches a much higher level of write IOPS.
Now let’s head to the next page, to look at the results
for the Myce/Oakgate Reads and Writes Tests…..