I've got a Windows client (desktop) application that when run for the first time will hit a server and request a client ID. A short time later that client will call back up to the server with additional requests. The server can hand out client IDs all day long, but it's important that the server be able to know that any future requests are coming from a valid client.
This means verifying that it's the client application making the request, and not someone/something else pretending to be the client application. Ideally it would also involve verifying that the client ID of a client application hasn't changed.
I thought about including a hash of the salted client ID with future requests, but it doesn't seem like it would be terribly hard for a determined person to disassemble the client enough to figure out the salt value.
Thanks in advance for any solutions, tips or pointers!
Can't be done sensibly without a heavy dose of cryptography. As long as the other end has complete control, there just is no way they can't fake the answers. Something vaguely similar is what Kerberos does, the protocols are quite involved. Ross Anderson's "Security Engineering" would be my first stop to take a look at how to do it.
Related
A little background: I am going to be constructing a webserver, likely the most up to date version of apache when I get around to it. It is going to be updated with sensory information from a makeshift security system I have.
As a counterpart, I am designing an app to go along with it, that will automatically contact the webserver and pull the sensory information about once every 1.5 minutes.
I want to have an authentication method so that the average Bob can't see this information, mostly due to the fact that there will be some command and control as part of the server as well.
The question: I feel like a simple username and password is the wrong way to go about this since it isn't dynamic and theoretically seeing the same credentials sent that frequent could be dangerous, so is there any other authentication method that could mitigate this?
The question pt. 2: Obviously I want an encrypted channel, will https stumble over itself if it tries to renegotiate every minute and a half?
I haven't begun this project yet much less chosen any language to write it in, meaning I am super open minded to suggestions, any help is greatly appreciated.
The question: I feel like a simple username and password is the wrong
way to go about this since it isn't dynamic and theoretically seeing
the same credentials sent that frequent could be dangerous, so is
there any other authentication method that could mitigate this?
You could use Google Sign-In to allow log on via a Google account.
Or you could implement two factor authentication with say Google Authenticator or via SMS to prove that the user logging in has more than one factor of authentication. These factors could be:
Something you know (e.g. password)
Something you have (e.g. phone that provides a One Time Password)
Edit: Having re-read your question - yes you are fine to authenticate with username and password (over HTTPS), however you should then store a session identifier client-side and simply send this in future rather than the username/password each time. This is more secure as it can be stored safely client-side, and if exposed the identifier can be easily revoked.
The question pt. 2: Obviously I want an encrypted channel, will https
stumble over itself if it tries to renegotiate every minute and a half?
Nope, this is what it is designed for. Browsers will keep open an HTTPS connection for a length of time. Additionally, they will use session resumption rather than executing a full HTTPS handshake in the case that a new connection needs to be established. Session resumption is much quicker than establishing a completely new session. See this article on the CloudFlare blog for more info.
I am thinking about writing a secure connection between a specific client and a specific server over HTTP. Of course SSL is the best and most obvious choice. But I keep thinking I could do the do the cryptography myself.
Note that this isn't about connecting any client to a specific server, but a specific client (e.g. a Java client app on my PC) to a specific server (my website hosted somewhere else). SO third-party certification doesnt seem necessary, since the server knows exactly which client to look out for and the cleint knows exactly which server to find.
If I want to upload a file from client to server. I could encrypt it manually (AES or other) and have the key hardcoded into the client app and also put in a file on the server, out of public view.
Please tell me if I'm crazy, stupid or pushing at windmills. Is my idea possible?
The short answer is, "No you can't." You can write some code that you think implements a secure connection, but actually it will be insecure. Designing and implementing a secure connection is a very skilled job; just see what a security consultant would charge you for doing that.
Since you are asking the question here, it is reasonable to assume that you are not such a person. Since you do not have the required skills and experience, anything you produce would be insecure.
Use existing standards where all the obvious errors, and a lot of the not so obvious ones, have been avoided for you.
Your idea possible, and really, if you just need to send file to specific URL, it would be easier to make a POST request with encrypted data (if URL/headers don't need to be secure for you).
Trying to roll your own replacement for SSL is a bad idea. Don't roll your own crypto. Instead, you should use SSL. It does what you need.
Given that you need to authenticate both the client and the server, you should use SSL with client certificates (as well as server certificates, which are standard).
Let's say I have 2 servers (server and authenticator), and I have a client. My end goal here is to be able to identify the client on server. My solution was to come up with a token/secret system like OAuth: client has a token and secret. It passes it to server. Server passes it to authenticator. If valid, server allows the request.
Obviously, this is nonoptimal just for the number of requests being made. The reason authenticator and server are separated is because this is for a decentralised service-- any number of servers may be used, and it's impractical to ask client libraries to register on each server.
So, the question remains, what's the best/correct way to do this? The goal is to create a system that is decentralised, but can still have clients identify themselves in a relatively secure fashion to the server.
Disclaimer: I'm not a security expert so I could be off-base here and in actual implementation there seems to be a number of security issues that would need to be ironed out.
In the broadest sense, could you have the client supply credentials to the authenticator and then upon verification the authenticator supplies the client and the server both with matching security tokens and then the client and server can communicate directly?
Just curious about there a reason you don't want to implement OAuth and run your own OAuth server.
Additional reference: http://groups.google.com/group/37signals-api/msg/aeb0c8bf67a224cc
Turns out the solution was to define my problem a bit better. As I'm only trying to create a way to block applications, I only need to store their name and key when they request the server. Then, as long as they're not blocked and the key matches the one in the datastore, they'll be identified. So I'm not trying to authenticate so much as identify. Thanks for the input!
I've done a little googling but have been a bit overwhelmed by the amount of information. Until now, I've been considering asking for a valid md5 hash for every API call but I realized that it wouldn't be a difficult task to hijack such a system. Would you guys be kind enough to provide me with a few links that might help me in my search? Thanks.
First, consider OAuth. It's somewhat of a standard for web-based APIs nowadays.
Second, some other potential resources -
A couple of decent blog entries:
http://blog.sonoasystems.com/detail/dont_roll_your_own_api_security_recommendations1/
http://blog.sonoasystems.com/detail/more_api_security_choices_oauth_ssl_saml_and_rolling_your_own/
A previous question:
Good approach for a web API token scheme?
I'd like to add some clarifying information to this question. The "use OAuth" answer is correct, but also loaded (given the spec is quite long and people who aren't familiar with it typically want to kill themselves after seeing it).
I wrote up a story-style tutorial on how to go from no security to HMAC-based security when designing a secure REST API here:
http://www.thebuzzmedia.com/designing-a-secure-rest-api-without-oauth-authentication/
This ends up being basically what is known as "2-legged OAuth"; because OAuth was originally intended to verifying client applications, the flow is 3-parts involving the authenticating service, the user staring at the screen and the service that wants to use the client's credentials.
2-legged OAuth (and what I outline in depth in that article) is intended for service APIs to authenticate between each other. For example, this is the approach Amazon Web Services uses for all their API calls.
The gist is that with any request over HTTP you have to consider the attack vector where some malicious man-in-the-middle is recording and replaying or changing your requests.
For example, you issue a POST to /user/create with name 'bob', well the man-in-the-middle can issue a POST to /user/delete with name 'bob' just to be nasty.
The client and server need some way to trust each other and the only way that can happen is via public/private keys.
You can't just pass the public/private keys back and forth NOR can you simply provide a unique token signed with the private key (which is typically what most people do and think that makes them safe), while that will identify the original request coming from the real client, it still leaves the arguments to the comment open to change.
For example, if I send:
/chargeCC?user=bob&amt=100.00&key=kjDSLKjdasdmiUDSkjh
where the key is my public key signed by my private key only a man-in-the-middle can intercept this call, and re-submit it to the server with an "amt" value of "10000.00" instead.
The key is that you have to include ALL the parameters you send in the hash calculation, so when the server gets it, it re-vets all the values by recalculating the same hash on its side.
REMINDER: Only the client and server know the private key.
This style of verification is called an "HMAC"; it is a checksum verifying the contents of the request.
Because hash generation is SO touchy and must be done EXACTLY the same on both the client and server in order to get the same hash, there are super-strict rules on exactly how all the values should be combined.
For example, these two lines provides VERY different hashes when you try and sign them with SHA-1:
/chargeCC&user=bob&amt=100
/chargeCC&amt=100&user=bob
A lot of the OAuth spec is spent describing that exact method of combination in excruciating detail, using terminology like "natural byte ordering" and other non-human-readable garbage.
It is important though, because if you get that combination of values wrong, the client and server cannot correctly vet each other's requests.
You also can't take shortcuts and just concatonate everything into a huge String, Amazon tried this with AWS Signature Version 1 and it turned out wrong.
I hope all of that helps, feel free to ask questions if you are stuck.
Lets say that in a browser based game, completing some action (for simplicity lets say someone clicks on a link that increases their score by 100) clicking on this link which would have a url for example increase_score.pl?amount=100 what kind of prevention is there from someone simply sending requests to the web server to execute this command:
Over and over again without actually doing the task of clicking on the link and
Sending a false request to the server where amount is set to something rediculus like 100000.
I am aware of checking HTTP_REFERER however I know people can get around that (not sure how exactly) and other than some bounds checking for the 2nd option I'm kind of stumped. Anyone ever experience similar problems? Solutions?
Nothing can stop them from doing this if you implement your game how you propose.
You need to implement game logic on the server and assign points only once the server validates the action.
For example: on SO when someone votes your question up, this isn't sent as a command to increase your reputation. The web-app just says to the server user X voted question Y up. The server then validates the data and assigns the points if everything checks out. (Not to say SO is a game, but the logic required is similar.)
Short version: you can't. Every piece of data you get from the client (browser) can be manually spoofed by somebody who knows what they're doing.
You need to fundamentally re-think how the application is structured. You need to code the server side of the app in such a way that it treats every piece of data coming from the client as a pack of filthy filthy lies until it can prove to itself that the data is, in fact, plausible. You need to avoid giving the server a mindset of "If the client tells me to do this, clearly it was allowed to tell me to do this."
WRONG WAY:
Client: Player Steve says to give Player Steve one gazillion points.
Server: Okay!
RIGHT WAY:
Client: Player Steve says to give Player Steve one gazillion points.
Server: Well, let me first check to see if Player Steve is, at this moment in time, allowed to give himself one gazillion points ... ah. He isn't. Please display this "Go Fsck Yourself, Cheater" message to Player Steve.
As for telling who's logged-in, that's a simple matter of handing the client a cookie with a damn-near-impossible-to-guess value that you keep track of on the server -- but I'll assume you know how to deal with session management. :-) (And if you don't, Google awaits.)
The logic of the game (application) should be based on the rule to not trust anything that comes from the user.
HTTP_REFERER can be spoofed with any web client.
Token with cookie/session.
You could make the link dynamic and have a hash that changed at the end of it. Verify that the hash is correct given that period of time.
This would vary in complexity depending on how often you allowed clicks.
A few things to note here.
First, your server requests for something like this should be POST, not GET. Only GET requests should be idempotent, and not doing so is actually a violation of the HTTP specification.
Secondly, what you're looking at here is the classic Client Trust Problem. You have to trust the client to send scores or other game-interval information to the server, but you don't want the client to send illegitimate data. Preventing disallowed actions is easy - but preventing foul-play data in an allowed action is much more problematic.
Ben S makes a great point about how you design the communication protocols between a client and a server like this. Allowing point values to be sent as trusted data is generally going to be a bad idea. It's preferable to indicate that an action took place, and let the server figoure out how many points should be assigned, if at all. But sometimes you can't get around that. Consider the scenario of a racing game. The client has to send the user's time and it can't be abstracted away into some other call like "completedLevelFour". So what do you do now?
The token approach that Ahmet and Dean suggest is sound - but it's not perfect. Firstly, the token still has to be transmitted to the client, which means it's discoverable by the potential attacker and could be used maliciously. Also, what if your game API needs to be stateless? That means session-based token authentication is out. And now you get into the deep, dark bowels of the Client Trust Problem.
There's very little you can do make it 100% foolproof. But you can make it very inconvenient to cheat. Consider Facebook's security model (every API request is signed). This is pretty good and requires the attacker to actually dig into your client side code before they can figure out how to spoof a reqeust.
Another approach is server replay. Like for a racing game, instead of just having a "time" value sent to the server, have checkpoints that also record time and send them all. Establish realistic minimums for each interval and verify on the server that all this data is within the established bounds.
Good luck!
It sounds like one component of your game would need request throttling. Basically, you keep track of how fast a particular client is accessing your site and you start to slow down your responses to that client when their rate exceeds what you think is reasonable. There are various levels of that, starting at the low-level IP filters up to something you handle in the web server. For instance, Stackoverflow has a bit in the web application that catches what it thinks are too many edits too close together. It redirects you to a captcha that you need to respond to if you want to continue.
As for the other bits, you should validate all input not just for its form (e.g. it's a number) but also that the value is reasonable (e.g. less than 100, or whatever). If you catch a client doing something funny, remember that. If you catch the same client doing something funny often, you can ban that client.
Expanding on Ahmet's response, every time they load a page, generate a random key. Store the key in the user session. Add the random key to every link, so that the new link to get those 100 points is:
increase_score.pl?amount=100&token=AF32Z90
When every link is clicked, check to make sure the token matches the one in the session, and then make a new key and store it in the session. One new random key for every time they make a request.
If they give you the wrong key, they're trying to reload a page.
I would suggest making a URL specific to each action. Something along the lines of:
/score/link_88_clicked/
/score/link_69_clicked/
/score/link_42_clicked/
Each of these links can do two things:
Mark in the session that the link has been clicked so that it wont track that link again.
Add to their score.
If you want the game to only run on your server, you can also detect where the signal is sent from in your recieving trick, and ignore anything not coming from your domain. It will be a real pain to tamper with your codes, if you have to run from your dedicated domain to submit scores.
This also blocks out most of CheatEngine's tricks.