Closed jerico closed 1 year ago
There's no notable change on the performance.
Maybe I'm missing something, but that's not how I'm reading the results; average response time went up from 0.87s to 1.19s, an increase of 0.32s/36%? That seems a pretty big difference though.
There's no notable change on the performance.
Maybe I'm missing something, but that's not how I'm reading the results; average response time went up from 0.87s to 1.19s, an increase of 0.32s/36%? That seems a pretty big difference though.
Sorry I averaged this in my head as "still around 1 second". That's where my comment came from.
Should I also do a test on how the original xhprof was performing? How should we know the threshold of acceptable range?
@rmccue I noticed that when I have another running task, it affects the response time.
I went ahead an re-ran the test. This time, more conscious of what other processes are running. For the following tests, I only have 1 instance of Altis project active, and no other apps aside from macOS system services running.
## 50ms
Transactions: 500 hits
Availability: 100.00 %
Elapsed time: 248.56 secs
Data transferred: 3.68 MB
Response time: 0.50 secs
Transaction rate: 2.01 trans/sec
Throughput: 0.01 MB/sec
Concurrency: 1.00
Successful transactions: 500
Failed transactions: 0
Longest transaction: 1.32
Shortest transaction: 0.41
## 5ms
Run #1
Transactions: 500 hits
Availability: 100.00 %
Elapsed time: 253.08 secs
Data transferred: 3.68 MB
Response time: 0.51 secs
Transaction rate: 1.98 trans/sec
Throughput: 0.01 MB/sec
Concurrency: 1.00
Successful transactions: 500
Failed transactions: 0
Longest transaction: 1.82
Shortest transaction: 0.42
Run #2
Transactions: 500 hits
Availability: 100.00 %
Elapsed time: 257.79 secs
Data transferred: 3.68 MB
Response time: 0.52 secs
Transaction rate: 1.94 trans/sec
Throughput: 0.01 MB/sec
Concurrency: 1.00
Successful transactions: 500
Failed transactions: 0
Longest transaction: 1.93
Shortest transaction: 0.42
Cool, that looks more like what I'd expect, no real impact to response times.
Merging despite linter check failing as there's an existing PR to fix those: https://github.com/humanmade/aws-xray/pull/79
Before we introduced excimer support, we were sampling at 5ms. PR to support excimer changed it to 50ms (see: https://github.com/humanmade/aws-xray/commit/8f8dcfed5b3cec2eef2ad2c157e34272448a193f#diff-a0f318343a79085d556c9f6050c1261041075f5129f00e908fce122d89e02febL24 )
This PR reverts back sampling interval to 5ms.
There's no notable change on the performance.
Making sure local xray is running:
Before
After
https://github.com/humanmade/aws-xray/issues/86