Basic Redis Security
In this article we’ll be going over the basics of Redis security. We’ll discuss the recommended model that Redis recommends as well as a few measures you can take to implement them.
Only Access with Trusted Clients
Redis was designed to be only be accessed by trusted clients. Only your servers running Redis should be able to access it. It should not be accessible to the internet at large.
What this means for a typical web app?
If you have a web application and the front-end needs some data that is stored in Redis, a request won’t be made directly to Redis, but to the application’s back-end which will in turn get the data from Redis. This prevents the browser from having direct access to Redis. In this type of implementation your backend can take some precautions such as validating the user input before making a request to Redis.
Protection by Authentication
You can setup authentication in the configuration file
redis.conf so that Redis only responds to queries from authorized users. Once you set this up a AUTH command with the password will be used to verify your credentials before your query is processed. The password that it must match is also stored in the
Redis itself does not support data encryption. If you’re using Redis correctly your Redis is being accessed locally and is not being sent over the wire and so you shouldn’t need encryption.
A common security measure in Redis is to rename commands that can do be used to do major damage. For example you could rename the CONFIG command to something obscure like this:
rename-command CONFIG 21591e49a59cfd7c
Now even if any hackers somehow gained access to Redis they still would not be able to use the CONFIG command without knowing it. You might want to protect the FLUSHALL command also because it deletes your entire data set.
Similarly to renaming you can disable a command by using the
rename-command and providing an empty string as the name. For example you could disable the CONFIG command like this:
rename-command CONFIG ""
Binding to Localhost
As we stated before the only servers accessing Redis should be the clients that are running Redis. So by default Redis is only accessible via localhose (127.0.0.1) but if you have changed this than you could have opened up a way for unauthorized users to gain access to your Redis instance.
To make sure your instance is bound to localhost you can find your
redis.conf file which is typicall in
/etc/redis/redis.conf and make sure that the following line is uncommented:
Attacks to Watch Out For
Watch out for DoS Attacks
Hackers can also use the system the way it’s intended to be used and still cause problems. If they know some form is putting all it’s data into Redis and there is no limit on the text put into the form, they can put large amounts of data into the form and overwork Redis which could eventually result in a slow service or a complete disruption.
Worst Case Attacks
There is a type of attacks that again tries to overwork Redis by giving it some worst case scenario data.
Redis doesn’t do string escaping so NoSQL injection is possible, although hopefully your backend is taking some precautions before sending requests to Redis.
Does it Lack Security?
Redis was built for speed, not safety. But if you have it setup up properly and you only have trusted clients accessing it then your data should be safe.
In this article we discussed the security model that Redis recommends. Remember that it should only accessed by trusted clients. We also discussed some types of attacks that Redis is susceptible to and some safety precautions you can take.
If you’re concerned about the security of your database please don’t hesitate to ask us. We’re experts at setting up and managing production data and we love it when people ask us about it.
Pilot the ObjectRocket Platform Free!
Try Fully-Managed CockroachDB, Elasticsearch, MongoDB, PostgreSQL (Beta) or Redis.Get Started