As you my or may not know, my day job is working for an online advertising company. The specific project I work on involves consuming XML formatted search feeds. We currently parse around 8,000 of these XML feeds per SECOND.
We spent months investigating the fastest way to parse XML. We eventually decided on a C library which walks the XML like a tree, it doesn’t load the XML DOM into memory like most conventional XML parsers do.
That all being said, XML is slowly dying across the internet. Most APIs are focusing on JSON interfaces instead of XML but for some reason our industry is just not willing to make the switch.