I understand that some applications need to check against a server periodically, but why don't servers let a mobile device know when it's time to update.
For example, let's say you receive an e-mail in gmail. As it stands now, there can be a delay between receiving the e-mail and your device notifying you. I assume this has to do with waiting until the device has checked the gmail server, at which poin开发者_高级运维t it sees the new e-mail and notifies the user. It makes more sense to me that when the gmail server receives an e-mail, it should ping (or however the creator wants to implement it) all registered devices to let them know that it's time to update. This way notifications would be synchronized and mobile device could save battery by not having to periodically check a server.
What you are describing is called a "server push". Some mobile applications does use server push. And Google actually have a technology in place to make this easier to implement, named "Cloud to device messaging" (C2DM). It works by having the servers sending a tiny little message, via googles C2DM servers, to the application running on the device, informing it its time for a update. And the device then retrieves the actual update from the original server (not from C2DM).
Its good stuff. It allows the service listening for C2DM manages to wake up the phone and everything. The downside is, the user of the application needs a google account. For android users, this isnt really a downside though.
Except they do, in some cases. In the "Mail, Contacts, Calendars" section of Settings on my iPod, there's a submenu labeled "Fetch New Data", which contains an option to have the mail server or whatever "Push" the data to the iPod. The Fetch aspect, which you can set the schedule for, is for when Push is off or you're syncing with applications that don't support push.
That is how Android does it. It pushes notifications to the device and that api is open to third party developers as well so they can push notifications to their users.
That thing is it is difficult to know the address of a mobile device at any particular time, the I.P address is dynamic, on a server the address is much more stable, so the device knows it will always find Google at google.com but Google has no real way of knowing where to find the device, it wouldn't know where to ping in the first place.
This is where push comes in, there are different implementations but i think they work by having the device maintain a connection to the server, and reestablishing it if its lost. See http://en.wikipedia.org/wiki/Push_technology
This of course is much harder to implement and maintain then a simple client pull.
精彩评论