New benchmarks in java #222
Conversation
…erless-benchmarks into java_benchmarks_support
| @@ -0,0 +1,33 @@ | |||
|
|
|||
There was a problem hiding this comment.
.vscode should be deleted :-)
| @@ -0,0 +1,10 @@ | |||
| { | |||
There was a problem hiding this comment.
This also should be deleted
| @@ -168,7 +168,7 @@ def build_base_image( | |||
| ) | |||
|
|
|||
| for fn in os.listdir(directory): | |||
| if fn not in ("index.js", "__main__.py", "Main.java"): | |||
| if fn not in ("index.js", "__main__.py"): | |||
There was a problem hiding this comment.
Why is the .java fil deleted now?
| stdout=subprocess.PIPE, | ||
| check=True, | ||
| ) | ||
| if code_package.language_name == 'java': |
There was a problem hiding this comment.
I'm wondering why we need to specify this argument for Java, and not for Python and Nodejs?
| ) | ||
|
|
||
| else: | ||
| subprocess.run( |
There was a problem hiding this comment.
A slightly less verbose version: just define the array of arguments for run, and only if language is Java, then append the two arguments of main.
| @@ -1,6 +1,6 @@ | |||
| { | |||
| "general": { | |||
| "docker_repository": "spcleth/serverless-benchmarks" | |||
There was a problem hiding this comment.
This is fine of course for testing but it shouldn't be committed :-)
| @@ -0,0 +1,38 @@ | |||
| { | |||
There was a problem hiding this comment.
Also shouldn't be commited
| @@ -5,67 +5,11 @@ | |||
| "update_storage": false, | |||
There was a problem hiding this comment.
Also should not be committed.
| jsonObject.put("compute_time", computeTime); | ||
| jsonObject.put("is_cold", isCold); | ||
| jsonObject.put("result", result); | ||
| jsonResult.addProperty("begin", formattedBegin); |
There was a problem hiding this comment.
I'm curious - what's the difference here :-)?
| @@ -0,0 +1,6 @@ | |||
| { | |||
| "timeout": 60, | |||
| "memory": 24, | |||
There was a problem hiding this comment.
this doesn't appear to be a correct value - is it really just 24 MB?
|
@mahlashrifi Few comments in addition to the review :-)
|
Hi,
I hope you are doing well. I have made a pull request to clarify which parts are unnecessary and which parts need to be added.
This pull request contains three components:
1- Completion of Java support for benchmarking on OpenWhisk (when the fat jar, including wrapper and source code, is less than 48 MB).
The initial commits are included in this pull request.
2- Two new benchmarks:
Training function of TeaStore
CheckLogin benchmark of TeaStore
3- Six additional benchmarks that are essentially variations of CheckLogin with different dependencies. I noticed something called "payload" in the results of running SeBS benchmarks, but I wasn’t sure how to use it. These six benchmarks were meant to serve as examples for using our implementation. My intention was to test whether the jar file size directly affects the delay in running serverless functions, but I found no significant impact.
The attached diagram and summarized information provide further details. I believe that if we can set the payload in SeBS, these additional codes can be omitted.