Monday, 8 August 2005
HTTP Performance (again)
Some folks at IONA have written a paper entitled Where HTTP Fails SOAP. I had a chance to look at this before I got it published, and their conclusions make a lot of sense — if you accept the premise that SOAP (and Web services) is about integration with existing applications.
In other words, if you’re using SOAP to integrate with existing systems, you’re probably going to use whatever interfaces are already available on the box; if there’s a HTTP server — say, Apache — you’ll use it, while if there’s a JMS queue, you might consider using that. If that’s the case, using HTTP will likely bring performance and resource problems; many widely-used HTTP servers (especially older implementations) aren’t that performant.
There’s a long way from that position, however, to one where this is all HTTP’s fault, and unfortunately that’s the impression that the article gives. As I’ve said before, there are very few fundamental limitations on performance in HTTP, and plenty of implementations that can easily match — or beat — those of other, “fast” protocols.
BTW, does anyone else find those sys-con.tv ads as annoying as I do?