Scrapy, find and fix a bottleneck: low throughput

READ DESCRIPTION: The scraper is based on Scrapy and it's almost complete, we are scraping products from AMZ (Italy at the moment) but we are struggling with overall requests throughout. We are using backconnect rotating proxies from 2 vendors and we have about 150 threads but we are able to scrape only 1/2 URLs per second, in addition we can use more threads/proxies with TOR but the speed is the same.