Pandan Azure masini na AWSu, 1vCPU, 1GB RAM, 20GB gp2 SSD storidza - 100IOPSa (tako pise u EBS - Volumes) - t2.micro. Azure masina ima 240 IOPSa po njihovoj konzoli. Odradio sam pre ovoga 7 run-ova da bih exhaustovao burst poene. Prilicno se sporo trose, a brzo se i skupljaju (jer kad sam bio burstovan, stvarno je cepalo ko ludo - zavrsavalo za ~300sek od 600max sa >2000IOPSa - sa ~2794941 operacija za to vreme):
burst poeni:
aws:
Code:
root@ip-172-31-35-64:~# sysbench --test=fileio --file-total-size=10G --file-test-mode=rndrw --max-time=600 --max-requests=0 run
sysbench 0.4.12: multi-threaded system evaluation benchmark
Running the test with following options:
Number of threads: 1
Extra file open flags: 0
128 files, 80Mb each
10Gb total file size
Block size 16Kb
Number of random requests for random IO: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Threads started!
Time limit exceeded, exiting...
Done.
Operations performed: 91740 Read, 61160 Write, 195679 Other = 348579 Total
Read 1.3998Gb Written 955.62Mb Total transferred 2.3331Gb (3.9817Mb/sec)
254.83 Requests/sec executed
Test execution summary:
total time: 600.0032s
total number of events: 152900
total time taken by event execution: 344.8785
per-request statistics:
min: 0.00ms
avg: 2.26ms
max: 90.10ms
approx. 95 percentile: 10.03ms
Threads fairness:
events (avg/stddev): 152900.0000/0.00
execution time (avg/stddev): 344.8785/0.00
root@ip-172-31-35-64:~#
Napravih ubuntu masinu a ni u jednom repou nema sysbench v1+, mrzelo me da ga kompajliram. Samo je matematika drugacija :p
Elem, i ovde sam dobio 250 IOPSa iako je bez bursta 100. Taman moze da se uporedi direktno sa Azureom jer mi on nudi 240 IOPSa. Ono sto je problem je sto masina i pored pada u IOPSu i dalje izvrsava skrejpovanje bez problema, dok Azure ne. Takodje instalacije preko yum-a u poredjenju sa CentOS-om na azureuu idu doooosta sporije nego na google cloudu - a ista regija je u pitanju i isti IOPS. Mislio sam da je to problem. Za poredjenje, evo iotesta i sa google cloud masinom:
gcloud:
Code:
Extra file open flags: 0
128 files, 80MiB each
10GiB total file size
Block size 16KiB
Number of IO requests: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Initializing worker threads...
Threads started!
File operations:
reads/s: 467.79
writes/s: 311.86
fsyncs/s: 997.94
Throughput:
read, MiB/s: 7.31
written, MiB/s: 4.87
General statistics:
total time: 600.2416s
total number of events: 1067046
Latency (ms):
min: 0.00
avg: 0.56
max: 333.94
95th percentile: 1.08
sum: 598268.25
Threads fairness:
events (avg/stddev): 1067046.0000/0.00
execution time (avg/stddev): 598.2682/0.00
[root@gcloud iotest]#
Google cloud je valjda 30 IOPSa po GB. VM ima 25GB.
Mislim da cu da pokusam da napravim novu Azure VM za pocetak. Osecam se kao da sam na nekom jeftinom openvzu gde nas 10 deli jedno jezgro.
THE ONLY EASY DAY WAS YESTERDAY