32 Comments

Summary:

I’m at a complete loss. Steve said at the Keynote that Yahoo would be providing a free push email account with every iPhone. But I don’t understand. Why is push email better than simply automatically checking the server every five seconds? Or, for that matter, every […]

I’m at a complete loss.

Steve said at the Keynote that Yahoo would be providing a free push email account with every iPhone.

But I don’t understand. Why is push email better than simply automatically checking the server every five seconds? Or, for that matter, every two minutes? Does it really matter if you get email, and then have to wait a whole two minutes before you get to see it?

Please explain this to me. Why does anyone care about push email?

  1. When the e-mail is pushed to your phone, you do not have to check regularly, thus minimizing data-traffic.

    With most telco’s in europe you have to pay for the amount of traffic you use (usually per megabyte).

    Share
  2. A POP request is a tiny amount of data; it surely couldn’t amount to a burden on the cell tower.
    I have a Sidekick that auto-fetches my email every few minutes; I too have wondered what the big deal about “push” is. I suppose there are times when you’d like an urgent email to pop right up, but email wasn’t really designed to do that, anyway. Lots of email servers queue messages for more than a few minutes.

    Share
  3. Rory Sinclair Monday, January 15, 2007

    Do you really need explained to you why an interrupt model is better than a polling model?

    Share
  4. Do you really have to use the nerdy technical names for the models?

    Share
  5. @A: A POP request may be a tiny amount of data, but like pennies, they can add up quickly. Imagine a phone that polls every 10 minutes (6 times per hour), now imagine 999 other phones in the same area that do the same — in one hour there will be 6000 POP requests, the vast majority of which will return negative results. The signal-to-noise ratio of POP requests is awful and they are often considered network garbage; wired netwroks may have the bandwidth to waste, but wireless networks of any type don’t. And remember, you’re still paying for sending that effectively useless data.

    In addition, push-email is safer for firewalled systems and many large companies require that any phones they purchase/lease have push-email support .

    Share
  6. One reason that push email is would be that people often have small “real-time” conversations over email (similar to SMS conversations). The user experience is just better when you know your email arrives almost instantly after you push the button.

    I’ve often had a phone conversation and had to email somebody a file (on regular email), and then continue the conversation based on the information in that file. I send it, the person in the other end says “yup, there it is”, and opens the file… It would be annoying if he’d have to wait for a couple of minutes for it to arrive. I know that email is not guaranteed (or supposed) to be instant, but in many cases these days, it is, and that makes it even better.

    Share
  7. Michael Houghton Monday, January 15, 2007

    From a system administrator’s point of view, it’s not just the pop ‘LIST’ command. It’s the fact that POP clients don’t stay connected, so there is a connection, an authentication hit on their radius (etc.) server, a list command (which lists everything, not just new messages), and a disconnect.

    Add in that most push services also do remote folder management and allow you to elect to get part of the message (e.g. the body but not the attachment) and the difference becomes obvious.

    That, and the fact that if someone mails you something, for example, while you’re on the phone to them, it appears as soon as it is available.

    And, do you really poll your POP server every five seconds? Your sysadmin must not be your biggest fan…

    Share
  8. I, presumably like Rory, was surprised to find that anybody would need to ask such a question… In the vast majority of cases (possibly all — I can’t think of an exception off the top of my head) a polling model is a terrible idea if a “notification” model can be used instead.

    If, as in the original post, one were to poll the server every 5 seconds, that would be 12 requests per minute, 720 per hour or 17,280 requests (as Michael laid out above, that includes connection, authentication, LIST request and disconnection) per day from just one phone. As Gruff pointed out, multiplying this by the number of phones connected to the network would result in an astronomical number of mostly redundant POP requests per day. Not only is this going to incur cost on the users’ behalf, but it would also slow the network to a crawl…

    It’s not a question of having your email 10 seconds earlier (although if you can, why not?), it’s a sensible way to decrease network traffic and reduce cost to everyone involved…

    Share
  9. Rory Sinclair Monday, January 15, 2007

    My apologies to Ben. Shall I rephrase using the terms ‘ask the network every 5 seconds’ and ‘the network tells you’?

    If you dont want to hear nerdy technical answers, dont ask nerdy technical questions.

    Share
  10. I just switched from a Windows Mobile device to a Blackberry. Polling my POP server every 1-5 minutes or so meant I’d get maybe half a work day out of the phone before the battery died – even with minimal voice usage.

    The Blackberry is getting well over two days of usage on a single charge. Email comes in the second it hits my server. Very slick.

    Share

Comments have been disabled for this post