Many companies now use PHP to write APIs to connect to the front-end web/iOS/Android. The cloud and end are completely separated, and dynamic negotiation cannot be achieved. If both parties agree on a static KEY, it will be transparent in the js of the web. Even ios/android can easily obtain this KEY through package interception or decompilation.
Nowadays, the maximum concurrency that general cloud services can accept is only a few thousand. If it is full by a denial of service attack, it will inevitably affect the access of normal users. Especially when using a communication protocol like websocket, once a connection is established, it will always be in trouble. If you just write a simple multi-line simulated connection to harass it, the server will basically fail.
How to properly prevent such attacks on the PHP side? There is no need to go too extreme. Controlling the paralysis within 10 minutes should still be acceptable for ordinary small applications.
Reply content:
In fact, the problem mentioned by the question is widespread. It can be said from two aspects:
Code aspect:
Programmer’s experience. Generally, it is the black and white list in terms of code, followed by the handling of interface timeout, and the rest are some specifications. In fact, there is no particularly good experience. More generally, ordinary hackers will not DDOS you directly. If it were a real DDOS, these would be enough for you. It is more about daily maintenance and handling of data anomalies.
Actually, you don’t have to worry about being maliciously attacked. To put it bluntly, the interface is also POST GET. As long as you filter the interface data to ensure the security of the data, common vulnerabilities can be prevented. Ordinary third-rate hackers will basically block this door. Unless your website is very famous, you can find a security company to do a test. Basically OK
To deal with DDoS attacks, it is more about system-level protection. The API interface written by PHP is generally for business. What PHP for business can do is not to hang up and respond as quickly as possible
In addition, the machine load capacity should be taken into consideration , basically this kind of problem requires the use of flow control to limit the number of times a terminal can access a certain API within a period of time.
1. The authentication interface receives a client time and a signature. First verify that the client time and server time are within a certain error range (such as ±5 minutes, the larger the error range, the less secure), and then verify whether the signature is consistent (the signature algorithm must use the client time). After success, a token and a random string.
2. All interfaces except the authentication interface must receive token and a signature. First check whether the token is legal, and then verify whether the signature is consistent (the signature algorithm must use the random string returned by the authentication interface).
3. Strategies for all interfaces:
1) If the signature verification fails a certain number of times (1 time, 2 times, 3 times, the larger the number, the less secure), the IP will be blocked.
2) If a single IP call exceeds the limit (it is best to set different restrictions for each interface according to business needs. For example, you can request 5 times/day for login), block the IP.
ip whitelist
If it happens too often, the IP will be blocked.
Don’t go to such trouble, use iptables