The browser wars don't interest me that much.  Personally, I want my browser to a) present my content in a pleasing intended way and b) not crash.  All the rest is gravy.  That said, I use both IE and Firefox, the former as it is best suited (or required) for some of the administrative tasks on my plate, and the latter because I prefer it.  So I downloaded Internet Explorer 8 this morning because I was curious about how well the touted gains in performance would bear out.  Before installing IE8, I tested Firefox 3 and IE7 against DevCentral to see the response times for loading the front page using HTTPWatch (and Wireshark to validate).  Then, I upgraded to IE8 and re-ran the tests. 

IE8 at first glance doesn't strike me as quite the advance that IE7 was at launch.  The first noticeable (and irritating) changes are that several of my add-on's don't work, as doesn't my VPN access, so there is a downgrade to IE7 in my immediate future...  There is a nice new tab feature, though.  If you launch one tab from another, they will be colored uniquely on the tab bar so you will have a visual queue to the related tabs.  Very nice, and useful when researching several unrelated things simultaneously.  OK, no more pleasantries, I upgraded for a little testing...


No real surprises here.  With IE7, the per-server connection limit was two.  It is now six with IE8, so triple the available connections should result in a more expedient response.  Firefox's connection limit is eight, though in all the test runs, no more than six were utilized (actually, both IE8 and Firefox uses six connections each to download the 56 objects from DevCentral).  Lori addressed the impact more connections will have on servers back in her post IE8: Robbing Peter to Pay Paul.  Basically, more connections equals more overhead, since the content amount doesn't change.  Great news for the user since their slice of the pie is very small, but not so great for the site administrators.  Anyway, this is about the user, not the site, so I digress.  You'll notice that except for an outlier in Firefox's MAX response, IE8 and Firefox are nearly identical in performance for both tests.

Cleared Cache



Since I was already in testing mode, I also wanted to see how their respective times would be with a primed cache courtesy of the features in the local browser and the WebAccelerator.  Again, performance is pretty even for IE8 and Firefox.

Primed Cache


Signification Reduction in Wait Time

Now that I had numbers with cleared cache and primed cache, I could see the benefit of the combination of the WebAccelerator and the caching features in the local browser.  You can gauge that gain by comparing the results of the two previous graphs, but the next graph summarizes the reduction in wait time as a percentage, which is pretty significant.  Cutting 70% of wait time from repeat visits is staggering.  Makes me wish I had a WebAccelerator in front of my browser to all my favorite sites.  Hmm....  Anyway, you can see that all three browsers show significant gains, but that IE7 gains the most, which makes sense since it has the least amount of serviceable connections per host available. 



In this quick and limited test, IE8 is inline with the performance I enjoy with Firefox.  All my tools don't yet support IE8, however, so I'll defer for now.


Follow me on Twitter Follow me on LinkedIn Follow me on Facebook