Logo

HDFS vs Ozone with Freon HCFS write test

Test method

  1. Start ozone freon dfsg -s1024000 -n10000 with -t 1 and -t 10 with both HDFS and Ozone path
  2. CRC calculation is disabled on both side
  3. Memdisk is used for persistent data on both side

Results

Time of execution (reported by the metrics of freon)

test HDFS Ozone
1 thread 186s, 181s 144s, 145s
10 threads 18s,19s 35s, 40s

https://github.com/elek/ozone-notes/tree/master/static/results/23_hcfs_write

Analysis

Ozone 10 thread test is measured with async-profile (with event=cpu and event=lock).

See:

lock profile shows time spent in Ratis SlidingWindow which contains a lot of syncrhonize-d method.

References

Ozone version: https://github.com/elek/hadoop-ozone/commit/bbf5d5e08a1c434cbe03ed05d4da912196196d0e

(upstream + HDDS-4285)

Deployment definition:

https://github.com/elek/ozone-perf-env/tree/master/hdfs-vs-ozone

Last commit:

https://github.com/elek/ozone-perf-env/commit/de841ca4cd3c6d1ceb56d0deee72844a0290a98f

Hardware

(physical machines, gru8)

VENDOR
	Manufacturer: Dell Inc.
	Product Name: PowerEdge R430
MEMORY
	Size: 32 GB
	Size: 32 GB
	Size: 32 GB
	Size: 32 GB
CPUs
	Version: Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz
	Max Speed: 4000 MHz
	Core Enabled: 10
	Thread Count: 20
	Version: Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz
	Max Speed: 4000 MHz
	Core Enabled: 10
	Thread Count: 20
DISKS
     NAME     SIZE TYPE FSTYPE MOUNTPOINT VENDOR   MODEL
     sda      1.8T disk                   ATA      ST2000NM0018-2F3
     ├─sda1 745.2G part xfs    /
     ├─sda2    10G part swap   [SWAP]
     └─sda3   1.1T part xfs    /var
     sdb      1.8T disk                   ATA      ST2000NM0018-2F3
     └─sdb1   1.8T part xfs    /data/1
     sdc      1.8T disk                   ATA      ST2000NM0018-2F3
     └─sdc1   1.8T part xfs    /data/2
     sdd      1.8T disk                   ATA      ST2000NM0018-2F3
     └─sdd1   1.8T part xfs    /data/3