Celestia Light node tasks started on march 29 and I am selected as one of the light node operators that I would like to give and compare some information about the hardware and Celestia software performance
For the Celestia Light node, recommended hardware requirements are given as ;
- Memory: 2 GB RAM
- CPU: Single Core
- Disk: 5 GB SSD Storage
- Bandwidth: 56 Kbps for Download/56 Kbps for Upload
I have run the Celestia Light node on Contabo Large VPS in the USA Region and the specs are like this;
My VPS Memory:
sudo lshq -short -C memory
My VPS CPU’s: 8 cores
lscpu
My SSD Disk result:
lsblk -d
df -H /
My VPS Bandwidth:
Sudo apt install speedtest-cli
speedtest-cli
COMPARISON
As shown below the machine specification is higher than recommended hardware given by Celestia Team and VPS only runs the Celestia light node alone. here is the comparison ;
- Memory: 2 GB RAM
- My VPS Memory: 30 GB RAM (Shared)
- CPU: Single Core
- My VPS CPU: has 8 Cores
- Disk: 5 GB SSD Storage
- My VPS Disk: 420GB SSD
- Bandwidth: 56 Kbps for Download/56 Kbps for Upload /
- My VPS Bandwidth: 468 Mbit/s Download/469.95 Mbit/s Upload
Operating System
cat /etc/os-release
#Output
NAME="Ubuntu"
VERSION="20.04.6 LTS (Focal Fossa)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 20.04.6 LTS"
VERSION_ID="20.04"
Celestia Light Node CPU performance
I have run the Celestia node as a service therefore I would like to measure the performance of the Celestia-lightd.service using the “perf” tool.
I used the following command to measure the memory usage of the Celestia-lightd.service:
sudo perf stat -p $(pidof celestia) sleep 10
This command measured the memory usage of the Celestia-lightd.service for 10 seconds and displays. I have tested the code for 10 / 30 / 60 seconds durations and the results are as;
root@vmi728168:~# sudo perf stat -p $(pidof celestia) sleep 10
Performance counter stats for process id '27760':
1,036.24 msec task-clock # 0.104 CPUs utilized
10,332 context-switches # 0.010 M/sec
914 cpu-migrations # 0.882 K/sec
0 page-faults # 0.000 K/sec
1,484,351,932 cycles # 1.432 GHz
238,914,034 stalled-cycles-frontend # 16.10% frontend cycles idle
394,956,947 stalled-cycles-backend # 26.61% backend cycles idle
1,066,790,002 instructions # 0.72 insn per cycle
# 0.37 stalled cycles per insn
79,746,473 branches # 76.957 M/sec
137,236 branch-misses # 0.17% of all branches
10.003413830 seconds time elapsed
root@vmi728168:~# sudo perf stat -p $(pidof celestia) sleep 30
Performance counter stats for process id '27760':
4,354.50 msec task-clock # 0.145 CPUs utilized
38,769 context-switches # 0.009 M/sec
4,450 cpu-migrations # 0.001 M/sec
78 page-faults # 0.018 K/sec
6,289,417,810 cycles # 1.444 GHz
1,267,687,538 stalled-cycles-frontend # 20.16% frontend cycles idle
1,176,377,297 stalled-cycles-backend # 18.70% backend cycles idle
2,446,005,529 instructions # 0.39 insn per cycle
# 0.52 stalled cycles per insn
277,048,160 branches # 63.623 M/sec
823,672 branch-misses # 0.30% of all branches
30.004314656 seconds time elapsed
root@vmi728168:~# sudo perf stat -p $(pidof celestia) sleep 60
Performance counter stats for process id '27760':
6,393.99 msec task-clock # 0.107 CPUs utilized
56,503 context-switches # 0.009 M/sec
5,773 cpu-migrations # 0.903 K/sec
162 page-faults # 0.025 K/sec
9,128,614,588 cycles # 1.428 GHz
1,610,033,688 stalled-cycles-frontend # 17.64% frontend cycles idle
2,049,734,788 stalled-cycles-backend # 22.45% backend cycles idle
5,433,829,680 instructions # 0.60 insn per cycle
# 0.38 stalled cycles per insn
457,107,718 branches # 71.490 M/sec
1,309,744 branch-misses # 0.29% of all branches
60.003704439 seconds time elapsed
In conclusion, My Contabo VPS is dedicated to running only The Celestia Light node and performs well without facing any problems.
https://tiascan.com/light-node/12D3KooWHh1bS5W9HUTu9DtQ9w4wUL5gFeqbRyMLbi6LGiHNxrMY