To use compression or not use compression.

that is the question

[twist]

I recently redesigned my site to be a little more purty. It was pretty plain and basic before but also had a really fast load time, about 4 seconds on 56k. The redesign has added a lot of weight, size is now ~70,000 bytes or about 15 seconds on a 56k.

I could turn on compression and probably knock about 5 or 6 seconds off that time. My problem is that I am on a shared server and anything I push on the server could come back and haunt me if the server is running slow. I have noticed this in the past. If the server is running slow while trying to compress, it could tack on 3 or 4 seconds to my time. This would affect broadband users, who are typically less patient with load times. A broadband user would load 70,000 bytes within a second, but with a slow shared server and compression, that same broadband user could end up waiting 4 or 5 seconds.

The other thing that I am wondering about is with all these dial-up companies offering built in high-speed compression through their services. Why don't I let them compress my pages for me, with no cost to me. Why force my server to compress a page that would have been compressed anyway down the line, possibly with even better, faster compression. It would seem that any dial-up user that cared about speed would have signed up for the high-speed dial-up anyway. If they don't, then they must like watching webpages load slowly.

Are the times of bending over backwards for 56k'ers (by compressing your content for those who wont pay for the high-speed dial-up) over?

[mcavic]

Are the times of bending over backwards for 56k'ers

Yes, I think so. On a shared server, CPU time is a big consideration. Also, if your design uses JPG files, they won't compress anyway.

[Farix]

Are the times of bending over backwards for 56k'ers (by compressing your content for those who wont pay for the high-speed dial-up) over?

No, because affordable broadband availability is still piss-poor in the US.

[twist]

No, because affordable broadband availability is still piss-poor in the US.

gzip compression is only for html text, before the server sends out the html it compresses it, at a cost of cpu time. Broadband users don't really require the compression of text as it's pretty irrelevant for them.

A lot, if not most all, dial-up companies offer a slightly higher priced service that will do this same compression before they send the pages to the user. So, now that most dial-up users can get the text compressed through their dial-up service, why should I waste cpu time and possible delays to 56k (who already have compression) and broadband users? The only benifit seems to be those who use dial-up and won't pay for the compression. The question is, should I even bother with those users?

(this referes to a shared server environment)

[tedster]

Why not try it and measure the results? Some sites find that it is a big help, and other not so much. And for some, like WebmasterWorld, there are technical factors that make gzip impractical [webmasterworld.com].

By the way, gzip compression iss not just helping the straight dial-up user. It also may help you getting spidered.

[twist]

That thread was what made me first try turning off gzip.

Why not try it and measure the results?

I know that my pages used to have about a 1 or 2 second initial delay when gzip was turned on that is now gone.

I don't have access to dial-up. I don't even know anybody anymore on dial-up. Unless I sign up for a free month of dial-up somewhere and buy a modem, I don't know how to see results over 56k. Any ideas?

[tedster]

By measure the results, I meant that in terms of how users relate to your site. Do you see more page views per unique? Do conversions go up? and so on. If there is a tangible benefit with compression , then it makes sense. If not, save the overhead.