Ross Wm. Rader wrote:
First: I'm only checking one place and getting everything. This is better (for me) then checking dozens or hundreds of places for something new.Slightly off topic, anybody else read any of the various "RSS is crushing my server" articles recently? There is a reason why infrequent or unpredictable updates should be pushed rather then pulled.How is the polling mechanism that most aggregators and NNTP clients use any different than the polling mechanism that most POP clients use?
It uses a fixed amount of resources to distribute a message of a size to a given number of users, plus a little protocol overhead. The amount of resources is roughly the same with a push or a pull technology.
However, every time I check your server and don't download a message, I've wasted both of our resources doing that check.
I was going to check mail anyway, so what's the difference if I check an RSS or NNTP server? Well, for one thing, it doesn't negate the need to check for mail. I check mail AND I am now checking something else that could have been sent by email.
Second: From both a server and client resource point of view, mail (POP3 especially, and IMAP to a lessor extent) only shows me new messages, I'm not constantly downloading headers or article overviews or anything else relating to messages I've already downloaded.
NNTP tells me the range of messages (two numbers, a high and a low) my client knows I've already read certain ranges and only downloads the rest. At most, I'll use LISTGROUP, which gives me the article numbers of every article in the group, but it won't download anything more.
With RSS I am downloading the entire list every time, my aggregator is parsing the list every time hoping to find something new. It wastes your time generating the list, it wastes both our bandwidth, and it wastes time hoping to find anything new.
This is an honest question - I'm having a hard time sorting out pull from push in this conversation because on my desktop, these are all request based transactions. i.e. "Whaddya got for me?"Another consideration: I am paying the people running the server I am accessing via POP3 (or IMAP, in my case) -- We have an SLA. If they tell me to back off, that I'm checking mail too often and their poor little server can't handle it, I vote with my dollars and move to a provider that wants my business. With RSS, if we can't come to terms, I can't just change the provider -- You're the only provider of that RSS feed.
In my case, I run my own server and I used to check mail every 60 seconds on most of my personal accounts, and every 90-300 seconds on my mail list accounts back when I used POP3.
I should also mention a couple other factors. IMAP doesn't require me to constantly build up and tear down TCP sessions, SSL handshake, reauthenticate, redownload the capabilities list and all the other housekeeping chores which are performed every time I connect to my server.
I keep five sessions per mailbox open for 8-12 hours a day. While those sessions are idling if anything new shows up in the mailbox, the server notifies my client. There is *no* polling going on at all unless mail actually shows up.
-- I'm sorry sir, you can't park your van on the diving board.
