suavesu

5 ms·

1.
Building a self-hosted cloud coding agent(stanislas.blog)
2 points by suavesu 7mo ago|1 comments
2.
by suavesu 7mo ago
Pause = delete VM. Resume = remount volume and keep going. Built a self-hosted cloud coding agent on k3s + Kata VMs, but the key trick is storage: JuiceFS (Redis metadata + S3 data + cache) makes sessions truly persistent.
3.
by suavesu 8mo ago
Some users use JuiceFS with CephFS RADOS, as alternative with Ceph MDS.
4.
by suavesu 8mo ago
pjdtest result of JuiceFS -> https://juicefs.com/en/blog/engineering/posix-compatibility-...
5.
by suavesu 8mo ago
Agree, don't recommend POSIX proxy on S3 for complex workload, like S3FS. In the design of JuiceFS, S3 is like raw disk, JuiceFS metadata engine is like partition table, compare with local file system.
6.
by suavesu 8mo ago
Can run MySQL and PG, but don't recommend, not good performance for production. but for temporary it's OK. Here's a case study -> https://juicefs.com/en/blog/user-stories/xiachufang-mysql-ba.
7.
by suavesu 8mo ago
metadata backup is very important. don't forget.
8.
by suavesu 8mo ago
The MLPerf performance of JuiceFS last year, FUSE client, and TCP/IP network. There're some analysis about performance and bottleneck. -> https://juicefs.com/en/blog/engineering/mlperf-storage-v2-
9.
by suavesu 8mo ago
Truth. It supports very large volume, more than 100PiB and 10B files in a single volume. Disclosure: I'm co-founder of JuiceFS
10.
by suavesu 8mo ago
yes, support more 10 options, include redis, SQL-like DB, TiKV, FoundationDB, and more. see here -> https://juicefs.com/docs/community/databases_for_metadata
11.
by suavesu 8mo ago
JuiceFS metadata engine comparison -> https://juicefs.com/docs/community/metadata_engines_benchmar...
12.
by suavesu 6y ago
I'm from JuiceFS team. We run JuiceFS full managed service on every public cloud worldwide for 4 years. We decide open source it for more explore and usage.