The gateway is not the exit
The first thing to keep straight is boring but important: your app connects to proxynade.net:2555. It does not chase residential IPs directly. The gateway picks a residential exit behind it, sends the request through that exit, and the target sees the ISP-side address.
The proxy line stays normal:
protocol://username:password@proxynade.net:2555Use HTTP, HTTPS, or SOCKS5 at the front. The gateway address and port stay the same. Your scraper can make ten requests through the same string and still land on ten different residential IPs because the gateway assignment changed, not your code.
That exit IP should look like a consumer ISP address. A French exit may show as Orange or Free. It should not look like a hosting ASN. That is why residential routes get used where datacenter routes look too much like servers. Datacenter is usually steadier and easier to debug, but it is also easier to classify by hosting ASN.
Residential has messier physics. Some exits are fast. Some are slow. Some drop in the middle of a sticky run because the underlying home connection changed. I have lost time blaming the gateway when the gateway was fine and the selected exit had simply gone bad.
This is the kind of trimmed log I mean. The original had target names and longer URLs:
13:42:08 w2 plan=premium proto=socks5 exit=orange-fr status=200 ms=812 bytes=18422 sticky=s7c9
13:42:11 w2 plan=premium proto=socks5 exit=orange-fr status=200 ms=904 bytes=19110 sticky=s7c9
13:42:15 w2 plan=premium proto=socks5 exit=orange-fr status=000 ms=30000 bytes=0 note=timeout sticky=s7c9
13:42:46 w2 plan=premium proto=socks5 exit=orange-fr status=000 ms=0 bytes=0 note=dead_exit sticky=s7c9
13:42:48 w2 plan=premium proto=socks5 exit=orange-fr status=200 ms=1041 bytes=18890 sticky=newRotating would usually move on faster because each request or connection can get a different exit. Sticky keeps the same IP for a duration, which helps when a site expects a small session to stay on one address. Hard sessions pin to one exit until that exit dies. Hard is useful when state matters. It is annoying when the exit disappears.
Premium and Volume use the same basic gateway idea. Volume Residential is $0.89/GB. Premium Residential is $5.00/GB. Both can target country, region, and city. Premium adds ISP, ASN, adblock, and stronger route quality. I only reach for those controls when the target cares about that precision. Otherwise Volume is the simpler high-volume lane.
What you control is the proxy line, protocol, plan, session mode, and targeting. What you do not control is the exact rotating exit, the speed of one home connection, whether that IP had a noisy morning before you got it, or how long it stays alive.
Usage logs are where I check the story. They report provider-metered bytes by host, plan, protocol, outcome, status, and latency. If bytes climb but saved rows do not, the problem may be target behavior, exit quality, retries, or session handling. The port number is rarely the interesting part.