I will be happy to get advice from more experienced developers about adding Web Sockets into my HTTP-based project.
That’s the thing. I have developed the REST API based service. Everything works well enough, but… In some special cases my server needs a long time to serve client requests. It may be from 1 minute to several hours (and even days)! I implement some not-so-good algorithm to address this issue:
Client sends HTTP request
Server replies about registering request
Client starts sending HTTP requests to get necessary data (if response does not have needed information the client sends another request and so on)
That is all in a nutshell.
And it seems to be a bad scenario and I am trying to integrate web sockets for adding duplex-channels in this architecture. I hope that my API will be able to send info about updated data as soon as possible without the necessity of many requests from the client.
But I am a bit confused in choosing one of two ways to use web socket (WS).
Variant A.
The server only tells the client via WS that data is ready. And the client gets data by standard request-response HTTP method from REST API.
Variant B.
The server sends all data to the client via WS without HTTP at all.
What variant is more suitable? Or maybe some other variants?
I do not want to remove HTTP at all. I just try to implement WS for a particular kind of end-points.
Variant A would be more suitable and easy to implement. You can send message to the client after the data is ready, and he can then send request for the data. It will be like a simple chat websocket, and will serve your purpose.
Related
For educational purposes I am trying to make a web api in c++. the web api needs to be able to listen for http requests(GET, POST etc.), when it receives a http request it needs to be able to send data back to the client. Because it is for educational purposes I would like to do it without unnecessary libraries. Now the first thing I need to do is make the api able to receive requests and respond on that, after some research on google I found out that winsock is probably the most basic way to setup sockets for windows but I could find very little on receiving http requests.
My question is: Is it possible with winsock to receive a http request from the browser, and send data back to the browser?.
My question is: Is it possible with winsock to receive a http request from the browser, and send data back to the browser?
Yes. ^^
It is, Because HTTP is a protocol that (usually) uses TCP as the underlying transportation protocol.
But trying to build a real HTTP layer on top of a simple win32 socket is a bit too much even for an experienced C++ developer.
Many un-experienced C++ developers would probably dismiss this task as a "well, you just need to read some data, parse the headers, assemble your own HTTP response and send it back".
but..
You will have to support
TLS, with all the nasty private keys/public keys implementation
Redirection
Chunked Transfer
G-Zip transfer
and the list goes on and on..
So practically speaking, if you just want to to accept a socket, read some data and send some basic HTTP response than yes. If you want a reliable, professional HTTP library - probably no.
You can check this page https://github.com/ReneNyffenegger/cpp-webserver to see simple winsock server implementation for HTTP. Web server implementation is not so difficult. Of course you should have time for it.
I am new to RESTful webservice. Whatever I have read over the internet about RESTful webservice, I came to know that REST works similar to servlet + webservice.
Our traditional webservice looks like JSP-> Servlet -> Service -> DAO -> Database.
Will REST replace Servlet in this heirarchy?
My ultimate goal is that my web application should support mobile application and normal browser also. Is it good idea to use REST in that case. If not, in what situation we should use REST?
I hope my question is clear.
Please help me.
Thanks in advance.
There are many ways we can achieve Machine to Machine communication.
Web services also helps communicating between applications made in different platforms.
For example a .net GUI can call a java server side program for data.
REST is one of that kind, based on HTTP protocol.
SOAP web service is heavy weight (using lots of XML) where as REST is simple and you can expose any of your APIS simply using REST.
A services exposed as REST services can be invoked by a client using on of the HTTP verbs GET, POST, PUT, and DELETE with their meaning same as in HTTP.
RESTful Web Services expose the state of its resources.
An 'Employee' data can be queried and represented in any format (Json, XML ...) using REST.
Rest won't replace the Servlet in your hierarchy, actually the HTTP based REST methods are written on this servlets.
Please go through this URL : http://docs.oracle.com/javaee/6/tutorial/doc/gijqy.html
Using REST is not related to browser experience on mobile or other devices. It totally depends on the client side technology used and your browser compatibility with those technologies.
Using REST is a good idea to access data at client side using simple AJAX calls.
REST means Representational State Transfer. It is a way of thinking about architecting network communication between client and server, with the focus being on transferring a resource from server to client and back again.
To understand the significance of this first consider a different architecture, Remote Procedure Call. This is where the client calls a function on the server as if the function existed on the client.
So you want to edit a photo that exists on the server. Your client is a photo editing app that uses RPC to achieve this. You want to blur the photo so your client calls the blur() function using RPC, and the server blurs the image and sends back the updated image. Then you want to rotate the image, so your client calls the rotate() function and the server rotates the image and sends the rotated image to your client.
You might have noticed two issues. Firstly, every time you carry out an action on the photo the server needs to do some work and send you back the updated image. This uses a lot of bandwidth.
Secondly what happens if tomorrow the server developers (who might be nothing to do with the client developers) decide that rotate() is the wrong function name, it should really be rotate_image(), and they update the server. Your client continues to call rotate() but this now fails because such a function doesn't exist on the client.
REST is an alternative way of thinking about client/server communication. Instead of telling the server to carry out an action on the resource (eg rotate the photo), why doesn't the client not just get a representation of the resource and carry out all the actions it wants to (blur, rotate etc) and then send the new state of the resource back to the server.
If you did it this way the protocol to communicate between client and server can be kept very simple and will require very few updates. All you need is functions for the client to get the resource and functions to put it back on the server. The client will have to know how to blur the image and rotate the image, but it doesn't need to know how to tell the server to do this, it just needs a way of telling the server to save the updated image.
This means that the developers of the client can work away implementing new features independently to the developers of the server. Very handy if the developers of the client are nothing to do with the server (the developers of Firefox have nothing to do with the New York Times website and vice versa)
HTTP is one such protocol that follows this architecture pattern and it allows the web to grow as it has. There are a small set of verbs (functions) in HTTP and they are concerned only with transferring a representation of the resource back and forth between client and server.
Using HTTP your photo client simply sends a GET message to the server to get the photo. The client can then do everything it wants to to the photo. When it is finished it sends the PUT message with the updated photo to the server.
Because there are not domain specific actions in the protocol (blur, rotate, resize) this protocol can also be used for any number of resources. HTTP doesn't care if the resource is a HTML document, a WAV file, a Javascript script, a PNG image. The client obviously cares because it needs to understand the resource it gets, and the server might care as well. But the protocol between the client and server doesn't need to care. The only thing HTTP knows is that there is a variable Content-Type in the HTTP header where the server can tell the client what type of resource this is.
This is powerful because it means you can update your client independently to updating your server without updating the transfer protocol. HTTP hasn't been updated in years. HTML on the other hand is updated constantly, and web servers and web browser are updated constantly (Chrome is on version 33). These updates can happen independently to each other because HTTP never (rarely) changes.
A web browser from 10 years ago can still communicate with a modern web server over HTTP to get a resource. The browser might not understand the resource, say it gets a WebM video that it can't understand, but it can still get this resource without the network communication failing.
Contrast that with the example of RPC above where the client server communication will break if the server changes rotate() to rotate_image(). Every single client will have to be updated with this new function or they will crash when trying to talk to the server.
So REST is a way of thinking about client server communication, it is an architecture design/pattern. HTTP is a protocol that works under this way of thinking that focuses on simply transferring state of a resource between server and client.
Now it is important to understand that historically a lot of people, including web developers, didn't get this. So you got things like developers putting verbs into resource names to try and simulate Remote Procedure Call over HTTP. Things like
GET http://www.mywebsite.com/image/blur_image
And they would hard code the URI /image/blur_image into their client and then try and make sure the guys developing the server never changed the URI blur_image. You get back to all problems of RPC. As soon as the server guys move the resource blur_image (which is not really a resource to start with) to /image/blur_my_image the client falls over because it has that hard coded as an action to perform, rather than simply getting /image and doing what ever it wants to it.
So there are lot of examples on the web of doing REST wrong. Anything that tightly couples client and server communication is doing REST wrong. Your client should be able to survive URIs changing, or Content-Types being updated, without falling over. It can complain it doesn't understand a resource (eg Netscape Navigator 2.0 complaining it has no idea what a HTML5 document is), but it should complain that a URI has changed. This is the discoverability aspect of REST, which I haven't gone into too much, but basically your client should be able to start at the root of the server http://www.mywebsite.com and if it understand the content types it should be able to continue on to the resource it wants. You should never need to hard code a URI into your client other than the root of the server.
I could write a book about this stuff (and many have), but I hope that serves as a good introduction about what REST actually is.
#javafan I just checked the mykong example you provided. Please note that that is not standard http servlet implementation, it is a Jersy way of implimentinmg rest. So when you map all your URIs goes through this servlet com.sun.jersey.spi.container.servlet.ServletContainer and you write classes with annotation #path etc the Jersy runtime environment will do the necessary processing for you like converting the input and output objects to necessary formats (json, xml etc) depending on your configuration. You can write a simple servlet and add methods in it with #path annotation in it and that will be invoked inturn when you make the corresponding request. but the doGet and doPost methods are standard servlet methods that processes GET and POST method by default. You can ad another methods to the same servlet and add more qualifiers to process your request.
#GET, #Produces("xml") etc.
I hope this helps.
Suppose that I have Apache/lighttpd or whatever to receive http requests. Now I want the web servers to act as a proxy for my web services running on internal servers written in Java/Clojure/Erlang?
What I want is to separate the layer that handles client connections and the server that handle application logic. These two should be separated and language independent. Is JSON or XML the format for communicating? If so, how do I perform it from the web servers?
Note: May be I missed the point of your question in this response. Pls do let me know if that is the case.
I dont think you should consider this as "forwarding" of the original request.
If your web-tier that receives the original request makes a call to one/more underlying services (thru HTTP or otherwise) it is part of the "processing" of the original request.
So, there is nothing different here than what you are already familiar with.
i.e You make a HTTP request in place where you would make a SOAP/XML request or a DB call or post a message.
When you say or think in terms of "forwarding", it is misleading.
Also, the data exchanged between your controller and services is solely based on your convenience.
It could be XML or JSON or regular POST parameters that gets sent over HTTP transport
How can I have two ways communication with WS. Two ways means a client could be a server and a server could be a client. As far as I understand the problem related to the client-server model in HTTP which is used by WS. What is the best practice for this scenario when a server wants send an event to multiple clients without being polled. ?
As far as I know there are some solutions but I do not know which one is best
1) server-push techniques (websockets)
2) SOAP over JMS (this sounds great)
3) WS-eventing
Thanks
The purpose of a webservice is - as the name says - to serve. It responds to requests, but it never sends requests on its own (but an application accessed through a webservice interface could send requests to other webservices to fulfill a request sent to it).
When a component of a service-oriented architecture is supposed to receive events from another component, it means that the receiving component has to act as a server and expose an own web services interface so that the component where the event occured can call it.
I am just getting started with SOAP web services and stumbled across WS-Addressing.
I have read the Wikipedia page, but I'm having a hard time understanding just what the point of WS-Addressing is.
According to Wikipedia and various sources on the web, WS-Addressing allows to put "addressing information" or "routing information" into the header of a SOAP request.
Why is this useful? If I send a request via HTTP (or even via SMTP or UDP), then the address I send to is the address of the server which will process my request, and the server can simply reply by the same channel. So why is addressing/routing information needed?
I'd be particularly interested in some real-world (more or less) example where WS-Addressing is helpful.
I've found WS-Addressing particularly useful in situations where the SOAP response can't be served immediately. Either the resources to form the response are not available right away or the result itself takes a long time to be generated.
That can happen when your business process involves "a human touch" for example (processes like the ones WS-HumanTask is targeting). You can stick web services in front of your business, but sometimes business takes time. It might be a subscription that must be manually verified, something to be approved, whatever, but it takes days to do it. Are you going to keep the connection opened all that time? Are you going to do nothing else than wait for the response? No! That is inefficient.
What you need is a notification process. The client makes a requests but does not wait for the response. It instead instructs the server where to send the response by use of a "reply to" address. Once the response is available, the server connects to that address and sends the response.
And voila... asynchronous interactions between web services, decoupling the lifetime of the communication process from the lifetime of the HTTP connection. Very useful...
But wait... HTTP connection? Why should I care about that? What if I want the response sent back on another type of protocol? (which SOAP kindly provides as it is not tied to any protocol).
With normal request/response flow, the response comes on the same channel as the request 'cose it's a connection you know.... So for example you have a HTTP connection... that means HTTP in and HTTP out.
But with WS-Addressing you are not tied to that. You can demand the response on another type of channel. Request comes on HTTP for example, but you can instruct the server to send the response back over SMTP, for example.
In this way WS-Addressing defines standard ways
to route a message over multiple transports. As the wiki page is saying:
instead of relying on network-level transport to convey routing information,
a message utilizing WS-Addressing may contain its own dispatch metadata in a standardized SOAP header.
and as for your observation:
and the server can simply reply by the same channel
... what works for some, might not work for others, and for others we have WS-Addressing :D.