So far, there is not yet completely free to developers and open Push service is available. There are several third-party push services abroad , but generally should be charged . So, in general , the domestic developers have to consider yourself to build Push services.
build their own Push service, a more natural choice is based on open source solutions do now .
use Google or Baidu search "Android Push Push " and other words , show that there are a lot of people have been studied . The top edge is such articles :
Android push way to achieve solutions
push to achieve with androidpn Push on to achieve
Android
Android Push Notification using information push to achieve
article mentioned above scheme, basically referred to an open source Android Push to achieve : androidpn .
androidpn it is essentially based on the server side Openfire , client -based asmack, which both open source implementation of the XMPP IM in the two basic components , it should be said ; androidpn just two more combine to make Push the scene .
I did chat App, here are willing to put practical experience doing open source XMPP -based IM system to others.
We do chat class App, more natural way , starting from the beginning is also open source XMPP IM systems research .
said first server-side options. Openfire is one of the oldest open source XMPP IM Server, do almost all the IM should have studied . However , it is also applied to the production of the most inopportune IM Server, because : single concurrency is very limited, cluster programs mature, old code and the lack of timely updates . For a concrete example : Openfire cluster assembly called Connection Manager , but you can see in the Openfire official website , the most recent version was released in February 2009 . Visible, based Openfire achieve androidpn is not enough to stabilize the foundation .
Update : App -based chat Openfire do communicate with a friend , their users than the larger, multiple Openfire cluster nodes do . Openfire they do a lot of transformation, such as XMPP protocol interaction complex , to simplify ; XMPP protocol text bloated , then converted to binary . Cluster is concerned, is entirely their own re- development. They load up to 300,000 users a single point .
In fact, relatively little has another two good options: ejabberd , tigase . ejabberd is using Erlang language, understand Erlang users rarely , it is generally not vote . We had a preliminary chat server option is tigase (Java implementation ) .
tigase authors maintain very active , the test results can be clustered to support a relatively large capacity, which is to attract our place. But after actual production operations point of view, because of its cluster program implementation complexity , as well as single- node capacity is limited, we do not have confidence in the 500,000 users on the cluster nodes to support , so the reach 500,000 before users , quickly developed its own alternative.
again for XMPP protocol with the client 's question: For mobile clients , the original XMPP somewhat complicated and traffic consumption . XMPP protocol is essentially the structure of the body are in the xml string , the amount of each agreement are a bunch of string , xml where there are a lot of meaningless structure. In addition , XMPP for its flexibility , you need to log in this matter there are N and forth . For mobile client care about traffic and capacity for , XMPP relatively bulky.
Our approach is : the agreement to a binary format , simplifying interaction on the agreement, but retained the original XMPP compatible .
androidpn Push is open source implementation is based on the XMPP open source components integrated, it does not make the necessary optimized for mobile application scenarios. In addition , XMPP IM is essentially a two-way agreement, and to achieve a direct XMPP -based Push function is not specifically optimized for Push features , such as client network strategies connections.
summarize in androidpn typical open-source Android Push program there will be a problem :
1) open source server implementation capacity stand , still need to improve their own open source implementation of new programs or completely re- use , high cost of investment and development is inevitable .
2) the agreement and implementation of such traffic consumption , network connectivity strategies , not specifically optimized for mobile Push is not economical.
our team based on the open source XMPP chat App system implementation experience , we have come to the conclusion that, in the IM client moving scenes , the open-source program is not an easy solution available . Later we own a complete re- architecture of the whole system . After that, this new architecture is based IM system , evolved out of the aurora push.
------ Solution ---------------------------------------- ----
androidpn have not been used, because there have been problems
------ Solution ----------------------- ---------------------
Aurora push push service is designed to do it
------ Solution ------- -------------------------------------
Aurora push push platform it is , then that is not their own development push slightly ---- - eference ---------------------------------------
analysis quite right ah
------ eference ------------------------------ ---------
look first ! ! ! ! ! ! ! ! ! !
------ eference --------------------------------------- < br> on Android push really not easy to achieve
------ eference ---------------------------- -----------
after watching the first horse
------ eference - --------------------------------------
is based on improved XMPP it
------ eference ---------------------------------------
their development is not a small cost , third-party platform can really save a lot of push things
------ eference ----------------- ----------------------
aurora push and principles is not the same as Google's C2DM it ?
------ eference --------------------------------------- < br> IOS platform so much easier , to achieve a simple push .
------ eference --------------------------------------- < br> XMPP protocol really is difficult to engage
------ eference ------------------------------- --------
learned , collection , thank you for sharing
------ eference -------------------- -------------------
learned the original androidpn is based on XMPP , ah
------ eference ----- ----------------------------------
aurora push with what agreement? Is xmpp it?
------ eference --------------------------------------- < br> Thank you for sharing
------ eference ----------------------------------- ----
look at third-party push
------ eference ---------------------------- -----------
Erlang small congregation of
------ eference --------------------- ------------------
support it, thank for sharing
------ eference ------------- --------------------------
professional third-party platform push it ?
------ eference --------------------------------------- < br> analysis was justified
------ eference ---------------------------------- -----
pit father GCM
------ eference ----------------------------- ----------
right, support open source
------ eference --------------------- ------------------
I am using XMPP for
------ eference -------------- -------------------------
MQTT also use ah ------ eference ------------------ ---------------------
openfire too old . . .
------ eference --------------------------------------- < br> friend Openfire transformation , and finally it is just a single point of capacity 300,000
------ eference ----------------------- ----------------
tigase authors maintain very active , the test results can be clustered to support a relatively large capacity
------ eference ----- ----------------------------------
for mobile clients , the original XMPP some complex and traffic consumption
------ eference ---------------------------------- -----
Openfire based foundation is not enough to achieve androidpn steady
------ eference ----------------- ----------------------
according to the landlord said , would be better to start from XMPP point ?
------ eference --------------------------------------- < br> with androidpn message push to achieve , such as the agreement with the realization flow consumption , network connectivity strategies , not specifically optimized for mobile push is not economical.
------ eference --------------------------------------- < br> open source server implementation capacity stand , still need to improve their own open source implementation
------ eference --------------------- ------------------
Android Push Notification realize I have seen the use of information push
------ eference - -------------------------------------
support a
------ eference - --------------------------------------
for mobile clients , the original XMPP traffic consumes some complex and large .
------ eference --------------------------------------- < br> XMPP to their flexibility , they log on this matter needs to have N and back . For mobile client care about traffic and capacity for , XMPP relatively bulky.
------ eference --------------------------------------- < br> tigase authors maintain very active , the test results can be clustered to support a relatively large capacity, which is to attract our place.
------ eference --------------------------------------- < br> we do chat class App, more natural way , but also from the study at the beginning of the open source XMPP IM system start.
------ eference --------------------------------------- < br> click to androidpn summarize the typical open source Android Push program will exist problems
------ eference ---------------- -----------------------
single concurrency is very limited, cluster programs mature, old code and the lack of timely updates .
没有评论:
发表评论