TrueCloudLab/frostfs-sdk-go

Optimize container nodes selection for the same netmap #86

New issue

Open

opened 2023-06-01 14:56:18 +00:00 by fyrchik · 2 comments

fyrchik commented

2023-06-01 14:56:18 +00:00

Owner

We have context, but it is private.
I think it makes sense to thing in 2 directions:

What info can we cache for different containers?
How can we modify selection/filtering process to reduce allocations?

This task is on the hot path (object put) and contributes a lot to the node memory consumption (policer).

We have `context`, but it is private. I think it makes sense to thing in 2 directions: 1. What info can we cache for different containers? 2. How can we modify selection/filtering process to reduce allocations? This task is on the hot path (object put) and contributes a lot to the node memory consumption (policer).

fyrchik added the

labels 2023-06-01 14:56:23 +00:00

ale64bit commented

2023-06-07 10:00:28 +00:00

Member

(...) a lot (...)

have we actually measured how much?

> (...) a lot (...) have we actually measured how much?

fyrchik commented

2023-07-11 15:13:43 +00:00

Author

Owner

Yes, this is based on the observation via pprof: under not-so-high load the most memory is allocated in policer.

Yes, this is based on the observation via pprof: under not-so-high load the most memory is allocated in policer.

Sign in to join this conversation.

No labels

good first issue

No milestone

No project

No assignees

2 participants

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: TrueCloudLab/frostfs-sdk-go#86

No description provided.