eds_speed_test: add GrpcMuxImpl config ingestion path to benchmark. #11566
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
A lot of the overhead we see in envoyproxy#11362 has to do with the overhead from the GrpcMuxImpl ingestion path, and helps put the EDS cluster overhead in context. This patch adds a GrpcMuxImpl and GrpcSubscription to the test, allowing a fairly realistic cost of EDS update to be obtained. Below is the example output. We're seeing v3 costs ~3.2x of the v2 ingestion. Future patches will narrow this. --------------------------------------------------------------------------------- Benchmark Time CPU Iterations --------------------------------------------------------------------------------- priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns 113 priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns 35 priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns 154 priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns 42 priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns 59 priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns 18 priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns 87 priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns 24 priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns 6 priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns 2 priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns 7 priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns 3 priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns 2 priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns 1 priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns 3 priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns 1 Risk level: Low (test only) Testing: Added ASSERTs to validate that we're loaded all the hosts at the end of the benchmark. Signed-off-by: Harvey Tuch <[email protected]>
|
CC @pgenera |
jmarantz
approved these changes
Jun 15, 2020
| : "type.googleapis.com/envoy.config.endpoint.v3.ClusterLoadAssignment"), | ||
| subscription_stats_(Config::Utility::generateStats(stats_)), | ||
| api_(Api::createApiForTest(stats_)), async_client_(new Grpc::MockAsyncClient()), | ||
| grpc_mux_(new Config::GrpcMuxImpl( |
There was a problem hiding this comment.
Sure, out of curiosity, what is the performance difference here?
There was a problem hiding this comment.
probably makes it slower if anything, but more conformant to modern style, no?
yashwant121
pushed a commit
to yashwant121/envoy
that referenced
this pull request
Jun 24, 2020
…nvoyproxy#11566) A lot of the overhead we see in envoyproxy#11362 has to do with the overhead from the GrpcMuxImpl ingestion path, and helps put the EDS cluster overhead in context. This patch adds a GrpcMuxImpl and GrpcSubscription to the test, allowing a fairly realistic cost of EDS update to be obtained. Below is the example output. We're seeing v3 costs ~3.2x of the v2 ingestion. Future patches will narrow this. --------------------------------------------------------------------------------- Benchmark Time CPU Iterations --------------------------------------------------------------------------------- priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns 113 priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns 35 priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns 154 priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns 42 priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns 59 priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns 18 priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns 87 priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns 24 priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns 6 priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns 2 priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns 7 priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns 3 priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns 2 priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns 1 priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns 3 priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns 1 Risk level: Low (test only) Testing: Added ASSERTs to validate that we're loaded all the hosts at the end of the benchmark. Signed-off-by: Harvey Tuch <[email protected]> Signed-off-by: yashwant121 <[email protected]>
yashwant121
pushed a commit
to yashwant121/envoy
that referenced
this pull request
Jun 24, 2020
…nvoyproxy#11566) A lot of the overhead we see in envoyproxy#11362 has to do with the overhead from the GrpcMuxImpl ingestion path, and helps put the EDS cluster overhead in context. This patch adds a GrpcMuxImpl and GrpcSubscription to the test, allowing a fairly realistic cost of EDS update to be obtained. Below is the example output. We're seeing v3 costs ~3.2x of the v2 ingestion. Future patches will narrow this. --------------------------------------------------------------------------------- Benchmark Time CPU Iterations --------------------------------------------------------------------------------- priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns 113 priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns 35 priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns 154 priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns 42 priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns 59 priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns 18 priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns 87 priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns 24 priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns 6 priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns 2 priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns 7 priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns 3 priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns 2 priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns 1 priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns 3 priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns 1 Risk level: Low (test only) Testing: Added ASSERTs to validate that we're loaded all the hosts at the end of the benchmark. Signed-off-by: Harvey Tuch <[email protected]> Signed-off-by: yashwant121 <[email protected]> Signed-off-by: yashwant121 <[email protected]> Signed-off-by: yashwant121 <[email protected]>
songhu
pushed a commit
to songhu/envoy
that referenced
this pull request
Jun 25, 2020
…nvoyproxy#11566) A lot of the overhead we see in envoyproxy#11362 has to do with the overhead from the GrpcMuxImpl ingestion path, and helps put the EDS cluster overhead in context. This patch adds a GrpcMuxImpl and GrpcSubscription to the test, allowing a fairly realistic cost of EDS update to be obtained. Below is the example output. We're seeing v3 costs ~3.2x of the v2 ingestion. Future patches will narrow this. --------------------------------------------------------------------------------- Benchmark Time CPU Iterations --------------------------------------------------------------------------------- priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns 113 priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns 35 priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns 154 priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns 42 priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns 59 priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns 18 priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns 87 priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns 24 priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns 6 priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns 2 priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns 7 priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns 3 priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns 2 priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns 1 priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns 3 priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns 1 Risk level: Low (test only) Testing: Added ASSERTs to validate that we're loaded all the hosts at the end of the benchmark. Signed-off-by: Harvey Tuch <[email protected]>
yashwant121
pushed a commit
to yashwant121/envoy
that referenced
this pull request
Jul 24, 2020
…nvoyproxy#11566) A lot of the overhead we see in envoyproxy#11362 has to do with the overhead from the GrpcMuxImpl ingestion path, and helps put the EDS cluster overhead in context. This patch adds a GrpcMuxImpl and GrpcSubscription to the test, allowing a fairly realistic cost of EDS update to be obtained. Below is the example output. We're seeing v3 costs ~3.2x of the v2 ingestion. Future patches will narrow this. --------------------------------------------------------------------------------- Benchmark Time CPU Iterations --------------------------------------------------------------------------------- priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns 113 priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns 35 priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns 154 priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns 42 priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns 59 priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns 18 priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns 87 priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns 24 priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns 6 priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns 2 priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns 7 priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns 3 priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns 2 priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns 1 priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns 3 priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns 1 Risk level: Low (test only) Testing: Added ASSERTs to validate that we're loaded all the hosts at the end of the benchmark. Signed-off-by: Harvey Tuch <[email protected]> Signed-off-by: yashwant121 <[email protected]>
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.
A lot of the overhead we see in
#11362 has to do with the
overhead from the GrpcMuxImpl ingestion path, and helps put the EDS
cluster overhead in context.
This patch adds a GrpcMuxImpl and GrpcSubscription to the test,
allowing a fairly realistic cost of EDS update to be obtained.
Below is the example output. We're seeing v3 costs ~3.2x of the v2
ingestion. Future patches will narrow this.
Benchmark Time CPU
Iterations
priorityAndLocalityWeighted/0/0/2000 6277063 ns 6276456 ns
113
priorityAndLocalityWeighted/1/0/2000 20236878 ns 20236505 ns
35
priorityAndLocalityWeighted/0/1/2000 4589757 ns 4588833 ns
154
priorityAndLocalityWeighted/1/1/2000 14853274 ns 14850006 ns
42
priorityAndLocalityWeighted/0/0/4096 11865838 ns 11865314 ns
59
priorityAndLocalityWeighted/1/0/4096 39787876 ns 39787716 ns
18
priorityAndLocalityWeighted/0/1/4096 8186874 ns 8185724 ns
87
priorityAndLocalityWeighted/1/1/4096 29376139 ns 29374980 ns
24
priorityAndLocalityWeighted/0/0/32768 110248743 ns 110246101 ns
6
priorityAndLocalityWeighted/1/0/32768 343826306 ns 343761359 ns
2
priorityAndLocalityWeighted/0/1/32768 76962915 ns 76953406 ns
7
priorityAndLocalityWeighted/1/1/32768 254426231 ns 254401890 ns
3
priorityAndLocalityWeighted/0/0/100000 352576987 ns 352550994 ns
2
priorityAndLocalityWeighted/1/0/100000 1100804152 ns 1100646263 ns
1
priorityAndLocalityWeighted/0/1/100000 253870688 ns 253857626 ns
3
priorityAndLocalityWeighted/1/1/100000 808080918 ns 808028852 ns
1
Risk level: Low (test only)
Testing: Added ASSERTs to validate that we're loaded all the hosts at
the end of the benchmark.
Signed-off-by: Harvey Tuch [email protected]