maskit commented on PR #13072:
URL: https://github.com/apache/trafficserver/pull/13072#issuecomment-4209901674

   ## JA4 performance impact                                                    
                                                                                
                                                                              
                                                                                
                                                                                
                                                                                
     Per fingerprint call (typical ClientHello with ~15 ciphers, ~15 
extensions):                                                                    
                                                                                
           
                                                                                
                                                                                
                                                                                
     | Allocation type | Before | After |                                       
                                                                                
                                                                                
     |---|---|---|   
     | Helper return strings (version, count, ALPN, hexify) | ~10 + 2×N 
ciphers/extensions | 0 |                                                        
                                                                                
        
     | Intermediate comma-separated string for hashing | 2 | 0 (streaming 
SHA256) |                                                                       
                                                                                
      
     | Vector copies for sorting | 2 | 0 (sort in place) |                      
                                                                                
                                                                                
     | ALPN string copy | 1 | 0 (string_view) |                                 
                                                                                
                                                                                
     | Result string | 1 | 0 (caller buffer) |                                  
                                                                                
                                                                                
                                                                                
                                                                                
                                                                                
     This eliminates 2–4 heap allocations and ~40 small-string 
construct/destruct cycles                                                       
                                                                                
                 
     per TLS connection. The per-connection savings are in the low-microsecond 
range —                                                                         
                                                                                
 
     modest compared to the TLS handshake itself, but meaningful in aggregate 
at high                                                                         
                                                                                
  
     connection rates (reduced allocator pressure and better cache utilization).


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to