Turso/fly.io: DB WRITE speed is seriously slow

It takes 0.5 seconds for 1 write to db on development PC. The problem seems to be in the network connection (routing). If I check using tracert, I am connecting to the Japanese db from Seoul, but it seems to go through Hong Kong and the US.

Is there any way to solve this problem?
I’ve tested several times over 2 days and always get the same result.

====
➜ ~ turso db show es-emoji
Name: es-emoji
URL: libsql://es-emoji-bahns.turso.io
ID: e1156573-198b-49dc-85a9-8329f84d537f
Group: default
Version: 0.24.30
Locations: ewr, fra, nrt
Size: 1.2 GB

Database Instances:
NAME TYPE LOCATION
nrt primary nrt
ewr replica ewr
fra replica fra

====
PS C:\Users\hello> tracert es-emoji-bahns.turso.io
es-emoji-bahns.turso.io 66.51.123.211둜 κ°€λŠ” 경둜 좔적:

1 <1 ms <1 ms <1 ms RT-AC1900P-3890 [192.168.1.1]
2 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
3 7 ms 4 ms 1 ms 10.19.238.141
4 2 ms 1 ms 1 ms 10.19.57.213
5 2 ms 2 ms 3 ms 1.213.15.25
6 2 ms 2 ms 2 ms 1.208.114.105
7 3 ms 2 ms 2 ms 1.208.113.41
8 2 ms 3 ms 1 ms 1.208.150.253
9 49 ms 49 ms 48 ms 1.208.145.42
10 46 ms 46 ms 46 ms 1.208.150.182
11 51 ms 49 ms 49 ms ae1-202.rt.iad.hkg.cn.retn.net [87.245.240.232]
12 222 ms 223 ms 221 ms ae0-6.rt.irx.sto.se.retn.net [87.245.233.172]
13 254 ms 253 ms 253 ms gw-as30081.retn.net [87.245.249.23]
14 257 ms 256 ms 256 ms 66.51.123.211

PS C:\Users\hello> nslookup es-emoji-bahns.turso.io 8.8.8.8
μ„œλ²„: dns.google
Address: 8.8.8.8

이름: es-emoji-bahns.turso.io
Addresses: 2a09:8280:1::6:d39d
66.51.123.211

===
My IP Address is:

IPv4: ? 182.231.xxxx
IPv6: ? Not detected
My IP Information:
ISP: LG Powercomm
City: Seoul
Region: Seoul-teukbyeolsi
Country: Korea (the Republic of)

======
7 Days later :

Same problem still exists. I’ve removed all other replicas except for the Japanese one, but requests are still too slow.

C:\Users\bahns>tracert es-emoji-bahns.turso.io

Tracing route to es-emoji-bahns.turso.io [66.51.123.211]
over a maximum of 30 hops:

1 <1 ms 2 ms <1 ms RT-AC1900P-3890 [192.168.1.1]
2 * * * Request timed out.
3 1 ms 3 ms 2 ms 10.19.238.141
4 2 ms 2 ms 2 ms 10.19.57.213
5 3 ms 2 ms 2 ms 1.213.15.25
6 2 ms 2 ms 2 ms 1.208.114.105
7 2 ms 2 ms 2 ms 1.208.113.41
8 30 ms 17 ms 3 ms 1.208.150.253
9 49 ms 48 ms 49 ms 1.208.145.42
10 45 ms 45 ms 45 ms 1.208.150.182
11 50 ms 48 ms 49 ms ae1-202.rt.iad.hkg.cn.retn.net [87.245.240.232]
12 223 ms 224 ms 223 ms ae0-6.rt.irx.sto.se.retn.net [87.245.233.172]
13 259 ms 261 ms 259 ms gw-as30081.retn.net [87.245.249.23]
14 270 ms 269 ms 269 ms 66.51.123.211

===

What should I do to solve this problem?

Test result : differences btw local vs. remote :

http://192.168.10.20:8080/
Temp table cleared successfully.
[Batch 100_1] Insert: 1.5005s (0.0150 s/rec)
[Batch 100_2] Insert: 1.5180s (0.0152 s/rec)
Total INSERT time: 3.0184s
Record per sec: 66.2593
Avg. time/rec: 0.0151 s/rec

Read 100 times, Avg read time: 0.014990s

libsql://es-emoji-bahns.turso.io
Temp table cleared successfully.
[Batch 100_1] Insert: 53.8226s (0.5382 s/rec)
[Batch 100_2] Insert: 53.3462s (0.5335 s/rec)
Total INSERT time: 107.1689s
Record per sec: 1.8662
Avg. time/rec: 0.5358 s/rec

Read 100 times, Avg read time: 0.669812s

--------------- Comparison Results ---------------
DB1 write is 35.50 times faster than DB2.
DB1 read is 44.69 times faster than DB2

Hello, this happens sometimes with anycast. I’ve forwarded the traceroute to our network providers and they’ll fix the issue (next week).

1 Like

we’ve applied a fix, is this looking better now?

It’s better now. But not good as before:
DB read 390ms/record is not production-ready result (tested on dev machine).
I’ll change my db region to HK area to fix this.
:

C:\Users\hello>tracert es-emoji-bahns.turso.io

μ΅œλŒ€ 30홉 μ΄μƒμ˜
es-emoji-bahns.turso.io 66.51.123.211둜 κ°€λŠ” 경둜 좔적:

1 <1 ms <1 ms <1 ms RT-AC1900P-3890 [192.168.1.1]
2 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
3 1 ms 2 ms 1 ms 10.19.238.141
4 2 ms 1 ms 1 ms 10.19.57.213
5 2 ms 2 ms 2 ms 1.213.15.25
6 1 ms 2 ms 2 ms 1.208.114.105
7 * * 2 ms 1.208.113.41
8 4 ms 4 ms 3 ms 1.208.150.253
9 37 ms 37 ms 37 ms 61.43.235.130
10 38 ms 37 ms 37 ms 61.43.231.78
11 45 ms 44 ms 45 ms 1.208.173.254
12 45 ms 45 ms 44 ms 203.208.154.13
13 80 ms 79 ms 80 ms 203.208.158.162
14 126 ms 127 ms 127 ms 203.208.143.138
15 125 ms 125 ms 125 ms 124.19.61.65
16 118 ms 118 ms 118 ms 59.154.58.74
17 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
18 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
19 168 ms 166 ms 165 ms 103.66.132.69
20 182 ms 182 ms 182 ms 66.51.123.211

===

libsql://es-emoji-bahns.turso.io
Temp table cleared successfully.
[Batch 100_1] Insert: 28.7266s (0.2873 s/rec)
[Batch 100_2] Insert: 29.3630s (0.2936 s/rec)
Total INSERT time: 58.0896s
Record per sec: 3.4430
Avg. time/rec: 0.2904 s/rec

Read 100 times, Avg read time: 0.393182s

--------------- Comparison Results ---------------
DB1 write is 16.65 times faster than DB2.
DB1 read is 22.49 times faster than DB2

It’s better now. But not good as before:
DB read 390ms/record is not production-ready result (tested on dev machine).
I’ll change my db region to HK area to fix this.
:
μ΅œλŒ€ 30홉 μ΄μƒμ˜
es-emoji-bahns.turso.io 66.51.123.211둜 κ°€λŠ” 경둜 좔적:

1 <1 ms <1 ms <1 ms RT-AC1900P-3890 [192.168.1.1]
2 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
3 1 ms 2 ms 1 ms 10.19.238.141
4 2 ms 1 ms 1 ms 10.19.57.213
5 2 ms 2 ms 2 ms 1.213.15.25
6 1 ms 2 ms 2 ms 1.208.114.105
7 * * 2 ms 1.208.113.41
8 4 ms 4 ms 3 ms 1.208.150.253
9 37 ms 37 ms 37 ms 61.43.235.130
10 38 ms 37 ms 37 ms 61.43.231.78
11 45 ms 44 ms 45 ms 1.208.173.254
12 45 ms 45 ms 44 ms 203.208.154.13
13 80 ms 79 ms 80 ms 203.208.158.162
14 126 ms 127 ms 127 ms 203.208.143.138
15 125 ms 125 ms 125 ms 124.19.61.65
16 118 ms 118 ms 118 ms 59.154.58.74
17 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
18 * * * μš”μ²­ μ‹œκ°„μ΄ λ§Œλ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€.
19 168 ms 166 ms 165 ms 103.66.132.69
20 182 ms 182 ms 182 ms 66.51.123.211