Improve Hubble memory usage and performance on decoding events #17482
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tklauser
added
kind/performance
|
/test |
tklauser
changed the title
rolinh
reviewed
Sep 28, 2021
kkourt
approved these changes
Sep 28, 2021
Found using staticcheck. Signed-off-by: Tobias Klauser <tobias@cilium.io>
To avoid regressions in successive changes. Signed-off-by: Tobias Klauser <tobias@cilium.io>
len is defined for all slices, even nil slices. These have a length of 0, so these checks are not necessary. Signed-off-by: Tobias Klauser <tobias@cilium.io>
Preallocate maps with known size and call json.Unmarshal directly, no need to allocate a json.Decoder. Suggested-by: Robin Hahling <robin.hahling@gw-computing.net> Signed-off-by: Tobias Klauser <tobias@cilium.io>
tklauser
force-pushed
the
hubble-mem-usage
branch
from
95f3f20
to
a2194ca
Compare
Found using prealloc. Signed-off-by: Tobias Klauser <tobias@cilium.io>
These are unused since commit b88afa2 ("monitor: new listener v1.2 with reusing en/decoder"). Signed-off-by: Tobias Klauser <tobias@cilium.io>
To avoid regressions in successive changes to DecodeTraceNotify. Signed-off-by: Tobias Klauser <tobias@cilium.io>
Use byteorder.Native.Uint16 to read the version uint16 directly instead of allocating a bytes.Reader. For L34Decode in hubble this improves performance and memory for each decoded TraceNotify event as follows: name old time/op new time/op delta L34Decode-8 4.30µs ± 7% 4.00µs ± 5% -6.96% (p=0.000 n=10+10) name old alloc/op new alloc/op delta L34Decode-8 1.24kB ± 0% 1.19kB ± 0% -3.87% (p=0.000 n=10+10) name old allocs/op new allocs/op delta L34Decode-8 26.0 ± 0% 23.0 ± 0% -11.54% (p=0.000 n=10+10) Signed-off-by: Tobias Klauser <tobias@cilium.io>
This avoids copying the Service protobuf object and improves performance and memory usage: name old time/op new time/op delta L34Decode-8 4.10µs ± 5% 3.63µs ± 8% -11.51% (p=0.000 n=10+10) name old alloc/op new alloc/op delta L34Decode-8 1.24kB ± 0% 1.03kB ± 0% -16.77% (p=0.000 n=10+10) name old allocs/op new allocs/op delta L34Decode-8 26.0 ± 0% 21.0 ± 0% -19.23% (p=0.000 n=10+10) name old time/op new time/op delta L7Decode-8 6.37µs ±11% 5.79µs ± 4% -9.02% (p=0.000 n=10+10) name old alloc/op new alloc/op delta L7Decode-8 1.53kB ± 0% 1.37kB ± 0% -10.47% (p=0.000 n=10+10) name old allocs/op new allocs/op delta L7Decode-8 36.0 ± 0% 34.0 ± 0% -5.56% (p=0.000 n=10+10) Also, the ok bool return is no longer necessary as callers can just check for a nil return value. This simplifies the logic in (*Parser).Decode a bit and avoids allocating an empty object for cases when no service is found. Signed-off-by: Tobias Klauser <tobias@cilium.io>
tklauser
force-pushed
the
hubble-mem-usage
branch
from
a2194ca
to
44e5ce3
Compare
|
/test |
aanm
approved these changes
Sep 28, 2021
rolinh
approved these changes
Sep 29, 2021
|
CI is green and reviews were provided by all teams. Marking as ready to merge. |
tklauser
added
the
ready-to-merge
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The first 7 commits are preparatory cleanup and addition/extension of tests to avoid regressions introduced by the successive two commits. These reduce Hubble's memory usage on L3/4 and L7 event decoding as follows:
See the individual commit messages for details.