Is using sessions in a RESTful API really violating RESTfulness? I have seen many opinions going either direction, but I'm not convinced that sessions are RESTless. From my point of view:
authentication is not prohibited for RESTfulness (otherwise there'd be little use in RESTful services)
authentication is done by sending an authentication token in the request, usually the header
this authentication token needs to be obtained somehow and may be revoked, in which case it needs to be renewed
the authentication token needs to be validated by the server (otherwise it wouldn't be authentication)
So how do sessions violate this?
client-side, sessions are realized using cookies
cookies are simply an extra HTTP header
a session cookie can be obtained and revoked at any time
session cookies can have an infinite life time if need be
the session id (authentication token) is validated server-side
As such, to the client, a session cookie is exactly the same as any other HTTP header based authentication mechanism, except that it uses the Cookie header instead of the Authorization or some other proprietary header. If there was no session attached to the cookie value server-side, why would that make a difference? The server side implementation does not need to concern the client as long as the server behaves RESTful. As such, cookies by themselves should not make an API RESTless, and sessions are simply cookies to the client.
Are my assumptions wrong? What makes session cookies RESTless?
First of all, REST is not a religion and should not be approached as such. While there are advantages to RESTful services, you should only follow the tenets of REST as far as they make sense for your application.
That said, authentication and client side state do not violate REST principles. While REST requires that state transitions be stateless, this is referring to the server itself. At the heart, all of REST is about documents. The idea behind statelessness is that the SERVER is stateless, not the clients. Any client issuing an identical request (same headers, cookies, URI, etc) should be taken to the same place in the application. If the website stored the current location of the user and managed navigation by updating this server side navigation variable, then REST would be violated. Another client with identical request information would be taken to a different location depending on the server-side state.
Google's web services are a fantastic example of a RESTful system. They require an authentication header with the user's authentication key to be passed upon every request. This does violate REST principles slightly, because the server is tracking the state of the authentication key. The state of this key must be maintained and it has some sort of expiration date/time after which it no longer grants access. However, as I mentioned at the top of my post, sacrifices must be made to allow an application to actually work. That said, authentication tokens must be stored in a way that allows all possible clients to continue granting access during their valid times. If one server is managing the state of the authentication key to the point that another load balanced server cannot take over fulfilling requests based on that key, you have started to really violate the principles of REST. Google's services ensure that, at any time, you can take an authentication token you were using on your phone against load balance server A and hit load balance server B from your desktop and still have access to the system and be directed to the same resources if the requests were identical.
What it all boils down to is that you need to make sure your authentication tokens are validated against a backing store of some sort (database, cache, whatever) to ensure that you preserve as many of the REST properties as possible.
I hope all of that made sense. You should also check out the Constraints section of the wikipedia article on Representational State Transfer if you haven't already. It is particularly enlightening with regard to what the tenets of REST are actually arguing for and why.
First, let's define some terms:
RESTful:
One can characterise applications conforming to the REST constraints
described in this section as "RESTful".[15] If a service violates any
of the required constraints, it cannot be considered RESTful.
according to wikipedia.
stateless constraint:
We next add a constraint to the client-server interaction:
communication must be stateless in nature, as in the
client-stateless-server (CSS) style of Section 3.4.3 (Figure 5-3),
such that each request from client to server must contain all of the
information necessary to understand the request, and cannot take
advantage of any stored context on the server. Session state is
therefore kept entirely on the client.
according to the Fielding dissertation.
So server side sessions violate the stateless constraint of REST, and so RESTfulness either.
As such, to the client, a session cookie is exactly the same as any
other HTTP header based authentication mechanism, except that it uses
the Cookie header instead of the Authorization or some other
proprietary header.
By session cookies you store the client state on the server and so your request has a context. Let's try to add a load balancer and another service instance to your system. In this case you have to share the sessions between the service instances. It is hard to maintain and extend such a system, so it scales badly...
In my opinion there is nothing wrong with cookies. The cookie technology is a client side storing mechanism in where the stored data is attached automatically to cookie headers by every request. I don't know of a REST constraint which has problem with that kind of technology. So there is no problem with the technology itself, the problem is with its usage. Fielding wrote a sub-section about why he thinks HTTP cookies are bad.
From my point of view:
authentication is not prohibited for RESTfulness (otherwise there'd be little use in RESTful services)
authentication is done by sending an authentication token in the request, usually the header
this authentication token needs to be obtained somehow and may be revoked, in which case it needs to be renewed
the authentication token needs to be validated by the server (otherwise it wouldn't be authentication)
Your point of view was pretty solid. The only problem was with the concept of creating authentication token on the server. You don't need that part. What you need is storing username and password on the client and send it with every request. You don't need more to do this than HTTP basic auth and an encrypted connection:
Figure 1. - Stateless authentication by trusted clients
You probably need an in-memory auth cache on server side to make things faster, since you have to authenticate every request.
Now this works pretty well by trusted clients written by you, but what about 3rd party clients? They cannot have the username and password and all the permissions of the users. So you have to store separately what permissions a 3rd party client can have by a specific user. So the client developers can register they 3rd party clients, and get an unique API key and the users can allow 3rd party clients to access some part of their permissions. Like reading the name and email address, or listing their friends, etc... After allowing a 3rd party client the server will generate an access token. These access token can be used by the 3rd party client to access the permissions granted by the user, like so:
Figure 2. - Stateless authentication by 3rd party clients
So the 3rd party client can get the access token from a trusted client (or directly from the user). After that it can send a valid request with the API key and access token. This is the most basic 3rd party auth mechanism. You can read more about the implementation details in the documentation of every 3rd party auth system, e.g. OAuth. Of course this can be more complex and more secure, for example you can sign the details of every single request on server side and send the signature along with the request, and so on... The actual solution depends on your application's need.
Cookies are not for authentication. Why reinvent a wheel? HTTP has well-designed authentication mechanisms. If we use cookies, we fall into using HTTP as a transport protocol only, thus we need to create our own signaling system, for example, to tell users that they supplied wrong authentication (using HTTP 401 would be incorrect as we probably wouldn't supply Www-Authenticate to a client, as HTTP specs require :) ). It should also be noted that Set-Cookie is only a recommendation for client. Its contents may be or may not be saved (for example, if cookies are disabled), while Authorization header is sent automatically on every request.
Another point is that, to obtain an authorization cookie, you'll probably want to supply your credentials somewhere first? If so, then wouldn't it be RESTless? Simple example:
You try GET /a without cookie
You get an authorization request somehow
You go and authorize somehow like POST /auth
You get Set-Cookie
You try GET /a with cookie. But does GET /a behave idempotently in this case?
To sum this up, I believe that if we access some resource and we need to authenticate, then we must authenticate on that same resource, not anywhere else.
Actually, RESTfulness only applies to RESOURCES, as indicated by a Universal Resource Identifier. So to even talk about things like headers, cookies, etc. in regards to REST is not really appropriate. REST can work over any protocol, even though it happens to be routinely done over HTTP.
The main determiner is this: if you send a REST call, which is a URI, then once the call makes it successfully to the server, does that URI return the same content, assuming no transitions have been performed (PUT, POST, DELETE)? This test would exclude errors or authentication requests being returned, because in that case, the request has not yet made it to the server, meaning the servlet or application that will return the document corresponding to the given URI.
Likewise, in the case of a POST or PUT, can you send a given URI/payload, and regardless of how many times you send the message, it will always update the same data, so that subsequent GETs will return a consistent result?
REST is about the application data, not about the low-level information required to get that data transferred about.
In the following blog post, Roy Fielding gave a nice summary of the whole REST idea:
http://groups.yahoo.com/neo/groups/rest-discuss/conversations/topics/5841
"A RESTful system progresses from one steady-state to the
next, and each such steady-state is both a potential start-state
and a potential end-state. I.e., a RESTful system is an unknown
number of components obeying a simple set of rules such that they
are always either at REST or transitioning from one RESTful
state to another RESTful state. Each state can be completely
understood by the representation(s) it contains and the set of
transitions that it provides, with the transitions limited to a
uniform set of actions to be understandable. The system may be
a complex state diagram, but each user agent is only able to see
one state at a time (the current steady-state) and thus each
state is simple and can be analyzed independently. A user, OTOH,
is able to create their own transitions at any time (e.g., enter
a URL, select a bookmark, open an editor, etc.)."
Going to the issue of authentication, whether it is accomplished through cookies or headers, as long as the information isn't part of the URI and POST payload, it really has nothing to do with REST at all. So, in regards to being stateless, we are talking about the application data only.
For example, as the user enters data into a GUI screen, the client is keeping track of what fields have been entered, which have not, any required fields that are missing etc. This is all CLIENT CONTEXT, and should not be sent or tracked by the server. What does get sent to the server is the complete set of fields that need to be modified in the IDENTIFIED resource (by the URI), such that a transition occurs in that resource from one RESTful state to another.
So, the client keeps track of what the user is doing, and only sends logically complete state transitions to the server.
As I understand, there are two types of state when we are talking about sessions
Client and Server Interaction State
Resource State
Stateless constraint here refers to the second type in Rest. Using cookies (or local storage) does not violate Rest since it is related to the first.
Fielding says: 'Each request from client to server must contain all of the information necessary to understand the request, and cannot take advantage of any stored context on the server. Session state is therefore kept entirely on the client.'
The thing here is that every request to be fulfilled on the server needs the all necessary data from the client. Then this is considered as stateless. And again, we're not talking about cookies here, we're talking about resources.
HTTP transaction, basic access authentication, is not suitable for RBAC, because basic access authentication uses the encrypted username:password every time to identify, while what is needed in RBAC is the Role the user wants to use for a specific call.
RBAC does not validate permissions on username, but on roles.
You could tric around to concatenate like this: usernameRole:password, but this is bad practice, and it is also inefficient because when a user has more roles, the authentication engine would need to test all roles in concatenation, and that every call again. This would destroy one of the biggest technical advantages of RBAC, namely a very quick authorization-test.
So that problem cannot be solved using basic access authentication.
To solve this problem, session-maintaining is necessary, and that seems, according to some answers, in contradiction with REST.
That is what I like about the answer that REST should not be treated as a religion. In complex business cases, in healthcare, for example, RBAC is absolutely common and necessary. And it would be a pity if they would not be allowed to use REST because all REST-tools designers would treat REST as a religion.
For me there are not many ways to maintain a session over HTTP. One can use cookies, with a sessionId, or a header with a sessionId.
If someone has another idea I will be glad to hear it.
i think token must include all the needed information encoded inside it, which makes authentication by validating the token and decoding the info
https://www.oauth.com/oauth2-servers/access-tokens/self-encoded-access-tokens/
No, using sessions does not necessarily violate RESTfulness. If you adhere to the REST precepts and constraints, then using sessions - to maintain state - will simply be superfluous. After all, RESTfulness requires that the server not maintain state.
Sessions are not RESTless
Do you mean that REST service for http-use only or I got smth wrong? Cookie-based session must be used only for own(!) http-based services! (It could be a problem to work with cookie, e.g. from Mobile/Console/Desktop/etc.)
if you provide RESTful service for 3d party developers, never use cookie-based session, use tokens instead to avoid the problems with security.
I am trying to create a Rest Api using a token based Authentication.
Is there any best practice in passing credentials from client to server for generating the token. As a HTTP Header or as a JSON String in post body?
I have been searching around but was not able to find any concrete answers.
Don't try to reinvent the wheel. For a good starting point look here:
best-practices-for-securing-a-rest-api-web-service
For my API implementation and my needs, I choose a simple BasicAuth (send credentials with the header) and any other tokens, and security related data I added to the JSON payload with each request.
Dont forget to set SSL as mandatory.
I would recommend using the Open ID Connect authentication protocol, and more specifically using a third party service or solid library that implements this protocol. Open ID Connect builds on OAuth 2 and is now widely used with support for various development languages and frameworks: http://openid.net/developers/libraries/
A successful authentication step results in an "access token" that can then be passed to your REST API where it is validated for authenticity. In Open ID Connect this token is passed as an HTTP header vs. the POST body.
If you roll your own protocol, or even develop your own Open ID Connect implementation, be aware of the details as it is very easy to overlook something small and thus create an insecure API. See the OAuth 2.0 Threat Model and Security Considerations for examples of what I am referring to. Due to this concern I always recommend use of an existing, well-vetted implementation.
I have a REST service that is already secured with basic authentication using the Authorization header. This is used to access the service in general and is required for any request. i.e. "User1", "password1".
I have a "file" resource which can have an additional password associated with it (i.e a password protected Word document, PDF, etc), "docpassword". What is the best way to send sensitive information like this? I'm especially interested in how to send the password for a GET request, but I'd like to have a universal solution that will also work for POST requests.
Maybe a custom header?
The HTTP protocol defines the standard Authorization header for sending authentication data (credentials) to the server. This header is defined in the RFC 7235 (which makes the old RFC 2616 obsolete and updates the RFC 2617):
4.2. Authorization
The Authorization header field allows a user agent to authenticate
itself with an origin server -- usually, but not necessarily, after
receiving a 401 (Unauthorized) response. Its value consists of
credentials containing the authentication information of the user
agent for the realm of the resource being requested.
Authorization = credentials
[...]
Please note that the name of this HTTP header is unfortunate because it carries authentication data instead of authorization. Anyways, this is the standard headers for sending credentials in the HTTP protocol.
Once you are already using the HTTP Basic Authentication Scheme to authenticate the users in your application, I believe you are already using the standard Authorization header.
I usually do not recommend custom headers, especially when the standard headers can be used instead, but your scenario seems not so common: You need to perform two authentications on the same request.
So, maybe a custom header such as X-Auth-Document or Document-Authentication with the document's password will be fine for GET requests. If you decide not using GET and decide using POST to access this resource, you can consider sending the document's password in the request paylod.
Anyways, don't forget using HTTPS: is highly advisable once you are sending sensitive data, such as credentials, over the wire. And HTTPS will protect you against the man-in-the-middle attack.
HTTP already has an authentication method, see for example this RFC: https://www.rfc-editor.org/rfc/rfc2617
Edit after clarification of question: There is nothing preventing the server to make additional challenges, even based on single resource basis. I must admit I did not implement such a thing yet, but each authorization can have its own realm. You could specify different realms, even down to document level if you really want to. The server then can potentially make multiple challenges in each realm (first login, then document). Remember, you can cache successful authentications, either on the client (for a realm, like a browser does), or giving out cookies with a cached token.
This would have the advantage of avoiding a custom header, and be completely HTTP/REST conform. There may be some performance disadvantage, but it could be mitigated with some targeted caching.
Of course you can go with custom header if you want to, but normally REST would imply that the client goes in with no prior knowledge other than mime-types and the HTTP protocol. A custom header implies out-of-band prior knowledge.
I'm writing REST API in PHP and recently I faced with authorization problem. I read a lot about basic authorization, about using private and public keys to create request signature. It is said that using request signature is more secure. But then I have a question:
-How should user will pass public key and generated signature?
I'm thinking about several options:
1) Create custom http header like X-Key, X-Signature
2) Use authorization header with custom scheme, like
AUTHORIZATION: SIGNATURE key='123' signature='abc'
3) Send this values as parameters. But I don't know if it acceptable for methods DELETE and PUT
What would you advice?
p.s. I don't want to implement oAuth
What are the desired properties of authentication scheme? Is this a publicly accessible or an intranet service? Are user accounts linked to something outside of scope of your API (linked 3rd party accounts etc). How are you going to distribute user credentials?
I would probably stick with plain old basic authorization, but encrypt everything at the transport level, making use of HTTPS mandatory. Rolling out your own cryptographic scheme is generally not a good idea. It's easier to fall victim to timing or replay attack than it seems. If you insist on client using a key pair for authentication, you can use HTTPS client certificates (though this is not widely used and maybe somewhat cumbersome solution).
There are a few security concerns about plain-text authentication over TLS. First, if someone implements MITM with forged certificate using either well known CA (maybe a government agency) or CA the client is forced to trust (big evil corporate proxy), they will get credentials. But you can't protect the client from its own environment anyway. Second, basic authentication can be prone to CSRF because browser knows how to do it and can remember credentials if you presented challenge and user filled the form. That's not a big problem if you adhere to REST principles and never allow state-changing GET requests. Also, if you are using JSON, never return arrays.
I'm building a picture diary on web application google app engine using python. Users can sign up and post pictures to their diary.
Also, I'm trying to conform as much as I can to the REST architecture of doing things.
The authentication scheme is based like this for the web application:
1. Post username/password from the frontend
2. Backend sets up a cookie if authentication is successful
3. The rest of the AJAX calls made are authenticated using this cookie.
Is there any way to conform to REST without using cookies ?
Now, I'm also building an android application where users can sign in and post/view their picture diary. I need to expose the data from web application's datastore so I'll be building a webservice to fetch data from the datastore.
The authentication scheme for the android client:
OPTION a
1. Post username/password over https to the web service
2. Web service returns a unique authorization token (store the token in the username/pwd table on the datastore)
3. Request subsequent services by adding this token to the Request Header of the request
4. Server maps the token to the username/pwd table and returns data if token is found
5. Authorization token expires after a certain period of time
OPTION b
1. Set up a secret key on the client and server side
2. Use "username:hash of password and secret key" in the authorization header of every request
3. server generates the password by extracting the password from the hash value using the same hash algorithm ; if successful returns data
btw, I didn't wanna use basic authorization because of its security vulnerabilities.
Which is better ?
Are there other significantly better ways to accomplish what I'm trying to do ? Security is quite a concern for me btw.
I'd appreciate if anyone has any insight into this issue. thanks.
I've been doing some research myself as to what would be the best solution. I think the 2-legged oauth might work in my case as Leonm suggested.
In this case the server has to provide the client with a consumer key/secret which in my case is hardcoded in the app.
The steps now would be:
1. Generate a signature using the oauth_parameters(consumer_key, signature_method, timestamp), request url, request parameters, and the SECRET.
2. Include the signature, oauth parameters when making a request.
3. Server verifies the request by generating the signature again except in this case it uses the SECRET that corresponds to the key
I think this way I am pretty much confirming to the REST principles. The server is statless as I far I understand now.
What are the pros/cons on doing things this way?
If "security is a concern" then I would say that you'd be a lot better off using open standards and a library to achieve what you want. The main reason for this is that if you do it yourself, you're very likely to forget something; these standards have had a lot of eyes looking at them, looking for holes.
Your options include (in increasing level of complexity)
Basic authentication and HTTPS
Everything is encrypted, which makes it impossible to compress or look into, it increases the overhead somewhat, using more horsepower on the server, and more perhaps battery power on the client. Simple to implement, since it's well supported by libraries.
Digest authentication
Unencrypted messages pass the wire, but the authentication is securely managed in the Authorization headers. See the wikipedia entry for more information.
OAuth
See how Google is providing OAuth for installed applications. I believe it isn't what you're looking for, since you're not asking to share data between applications, just authenticating users.
Roll your own
If you want to roll your own, I suggest looking at e.g. how Google's (now deprecated ?) ClientLogin used to work.
Clients would GET a protected resource, and get a 401 with instructions to perform a GoogleLogin authentication, including a URI for where to perform the login itself
Clients (knowing how to do this) POST a request in a specific manner to that URI
The server responds with a specific response including a (long) token
The client can now perform GET requests to the protected resource with that token.
Statelessness
You cite REST, which dictates that requests should not specifically depend on prior interaction: "... each request from client to server must contain all of the information necessary to understand the request, and cannot take advantage of any stored context on the server." (fielding) This means that a server shouldn't store conversational context (like an authentication token) in a table.
One way of fixing this is by using any of the token based approaches (where the server tells the client about a token it should use for future requests) where the token is not a random number, but a message to the server itself. To protect yourself from client tampering, it can be signed, and if you're afraid of clients looking at it, you can encrypt it.
Edit: Although I'm not certain, it seems unlikely that Google has a table of all authentication tokens ever issued; The length of their tokens suggests that the token is some encrypted message proving that whoever holds this token actually provided real credentials in some realm at some time.
OAuth does exactly what you want to do in a standard way.
You could use a combination of HTTPS and HTTP Basic Auth. Both are existing standards and should be secure enough when used together.