You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are building autocompletion using nodejs, sphinx and trying out limestone.
Autocompletion containts only 6 items
Everything works perfect except one performance issue.
When searching we use limit: 6, so we need quite small amount of data to return.
Sphinx returns results in average time from 1 - 9ms
However with header Transfer-Encoding': 'chunked. it takes browser 180-200 ms to receive data. And it is huge time loss for us.
If i turn off chunked header, then browser recieved data in 10-16ms, which id perfect and desired result, but the response is not complete if headers ar not specified as chunked.
If i take full response data as string from chunked result and send it directly as body, then it also takes 10-16ms. So the problem is definately in that chunked data output or buffering.
Is there any way to output straight and not chunked data for small amounts of items? Otherwise these 180-200ms are huge performance loss for us.
Thanks
The text was updated successfully, but these errors were encountered:
If anyone is interested here is how i got performance i wanted, avoiding chunked response. Now it takes maximum 10-15ms until request is received by browser:
Set chunkedEncoding= false, so that it don't use chunked encoding
We are building autocompletion using nodejs, sphinx and trying out limestone.
Autocompletion containts only 6 items
Everything works perfect except one performance issue.
When searching we use limit: 6, so we need quite small amount of data to return.
Sphinx returns results in average time from 1 - 9ms
However with header Transfer-Encoding': 'chunked. it takes browser 180-200 ms to receive data. And it is huge time loss for us.
If i turn off chunked header, then browser recieved data in 10-16ms, which id perfect and desired result, but the response is not complete if headers ar not specified as chunked.
If i take full response data as string from chunked result and send it directly as body, then it also takes 10-16ms. So the problem is definately in that chunked data output or buffering.
Is there any way to output straight and not chunked data for small amounts of items? Otherwise these 180-200ms are huge performance loss for us.
Thanks
The text was updated successfully, but these errors were encountered: