mirror of
https://github.com/deepseek-ai/3FS.git
synced 2026-11-05 00:47:47 +00:00
[GH-ISSUE #112] stripeSize概念疑问 #66
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
deepseek-ai/3FS#66
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @hmy1229 on GitHub (Mar 6, 2025).
Original GitHub issue: https://github.com/deepseek-ai/3FS/issues/112
每个chunk直接对chainTable中的chainNum直接取余即可得到chain:
chain = chains[chunkIndex % chainsNum]
而且这样能将文件打的更散
为何需要新增stripeSize来控制能打散到的chain的数量,有什么好处?增加这个stripeSize不就更不容易打散了么?
@beef9999 commented on GitHub (Mar 7, 2025):
Suppose you have 10,000 machines and an extremely large file. If you fully scatter the file to all chains, then when reading this file, your client would have to establish RDMA connections with all 10,000 machines. Not to mention the impact of long-tail latency caused by machine failures. Therefore, it is a better choice to limit the range the scattering.
假如你有10000台机器,和一个特别大的文件,如果是全打散,那么读这个文件的时候,你的客户端就得跟10000台机器建立RDMA链接。更别说宕机导致的长尾延迟问题。所以控制打散范围是更好的选择。