ivampiresp/Spdk
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

bdevperf: IOPS average by exponential moving average

Browse Source 
- Cumulative Moving Average (CMA): average of all data up to current
- Exponential Moving Average (EMA): weighted mean of the previous n
  data and more weight is given to recent
- Simple Moving Average (SMA): unweighted mean of the previous n data

Currently bdevperf support only CMA to know average of IOPS. However
CMA may not be appropriate to know the trend of IOPS. EMA or SMA will
be better to take the role.

The formula of EMA utilized in bdevperf is the following

Multiplier = (2 / (Time periods + 1))
EMA: EMA[n+1] = (IOPS - EMA[n]} x multiplier + EMA[n].

One of the source of the formula is https://github.com/patharanordev/ema

Change-Id: Ia7a282f6d173ace470e05b19ee6dee5edf640617
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/398886
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This commit is contained in:
Shuhei Matsumoto 2018-03-06 09:13:05 +09:00 committed by Jim Harris
parent 4492fd6cc8
commit 1732428a2b
1 changed files with 55 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

63
test/lib/bdev/bdevperf/bdevperf.c
View File

@ -64,6 +64,7 @@ static uint64_t g_time_in_usec;
static int g_show_performance_real_time = 0;
static uint64_t g_show_performance_period_in_usec = 1000000;
static uint64_t g_show_performance_period_num = 0;
static uint64_t g_show_performance_ema_period = 0;
static bool g_run_failed = false;
static bool g_shutdown = false;
static uint64_t g_shutdown_tsc;
@ -85,6 +86,8 @@ struct io_target {
struct io_target *next;
unsigned lcore;
uint64_t io_completed;
uint64_t prev_io_completed;
double ema_io_per_second;
int current_queue_depth;
uint64_t size_in_ios;
uint64_t offset_in_ios;
@ -570,11 +573,43 @@ static void usage(char *program_name)
printf("\t\t(read, write, randread, randwrite, rw, randrw, verify, reset)]\n");
printf("\t[-M rwmixread (100 for reads, 0 for writes)]\n");
printf("\t[-t time in seconds]\n");
printf("\t[-P Number of moving average period]\n");
printf("\t\t(If set to n, show weighted mean of the previous n IO/s in real time)\n");
printf("\t\t(Formula: M = 2 / (n + 1), EMA[i+1] = IO/s * M + (1 - M) * EMA[i])\n");
printf("\t\t(only valid with -S)\n");
printf("\t[-S Show performance result in real time in seconds]\n");
}
/*
* Cumulative Moving Average (CMA): average of all data up to current
* Exponential Moving Average (EMA): weighted mean of the previous n data and more weight is given to recent
* Simple Moving Average (SMA): unweighted mean of the previous n data
*
* Bdevperf supports CMA and EMA.
*/
static double
get_cma_io_per_second(struct io_target *target, uint64_t io_time_in_usec)
{
return (double)target->io_completed * 1000000 / io_time_in_usec;
}
static double
get_ema_io_per_second(struct io_target *target, uint64_t ema_period)
{
double io_completed, io_per_second;
io_completed = target->io_completed;
io_per_second = (double)(io_completed - target->prev_io_completed) * 1000000
/ g_show_performance_period_in_usec;
target->prev_io_completed = io_completed;
target->ema_io_per_second += (io_per_second - target->ema_io_per_second) * 2
/ (ema_period + 1);
return target->ema_io_per_second;
}
static void
performance_dump(uint64_t io_time_in_usec)
performance_dump(uint64_t io_time_in_usec, uint64_t ema_period)
{
uint32_t index;
unsigned lcore_id;
@ -591,10 +626,12 @@ performance_dump(uint64_t io_time_in_usec)
printf("\r Logical core: %u\n", lcore_id);
}
while (target != NULL) {
io_per_second = (double)target->io_completed * 1000000 /
io_time_in_usec;
mb_per_second = io_per_second * g_io_size /
(1024 * 1024);
if (ema_period == 0) {
io_per_second = get_cma_io_per_second(target, io_time_in_usec);
} else {
io_per_second = get_ema_io_per_second(target, ema_period);
}
mb_per_second = io_per_second * g_io_size / (1024 * 1024);
printf("\r %-20s: %10.2f IO/s %10.2f MB/s\n",
target->name, io_per_second, mb_per_second);
total_io_per_second += io_per_second;
@ -614,7 +651,8 @@ static void
performance_statistics_thread(void *arg)
{
g_show_performance_period_num++;
performance_dump(g_show_performance_period_num * g_show_performance_period_in_usec);
performance_dump(g_show_performance_period_num * g_show_performance_period_in_usec,
g_show_performance_ema_period);
}
static int
@ -772,7 +810,7 @@ main(int argc, char **argv)
mix_specified = false;
core_mask = NULL;
while ((op = getopt(argc, argv, "c:d:m:q:s:t:w:M:S:")) != -1) {
while ((op = getopt(argc, argv, "c:d:m:q:s:t:w:M:P:S:")) != -1) {
switch (op) {
case 'c':
config_file = optarg;
@ -799,6 +837,9 @@ main(int argc, char **argv)
g_rw_percentage = atoi(optarg);
mix_specified = true;
break;
case 'P':
g_show_performance_ema_period = atoi(optarg);
break;
case 'S':
g_show_performance_real_time = 1;
show_performance_period_in_usec = atoi(optarg) * 1000000;
@ -833,6 +874,12 @@ main(int argc, char **argv)
}
g_time_in_usec = time_in_sec * 1000000LL;
if (g_show_performance_ema_period > 0 &&
g_show_performance_real_time == 0) {
fprintf(stderr, "-P option must be specified with -S option\n");
exit(1);
}
if (strcmp(workload_type, "read") &&
strcmp(workload_type, "write") &&
strcmp(workload_type, "randread") &&
@ -940,7 +987,7 @@ main(int argc, char **argv)
}
if (g_time_in_usec) {
performance_dump(g_time_in_usec);
performance_dump(g_time_in_usec, 0);
} else {
printf("Test time less than one microsecond, no performance data will be shown\n");
}