Encrypt SQLite database without increasing app size?
The increase of 10MB in your APK when using SqlCipher can be avoided by shipping a separate APK for each Android architecture you want to target.
For performance and cross platform support reasons SqlCipher is implemented in C code and built as a native Android .so library using the Android NDK. A different library is built for each Android architecture it supports, which in the case of SqlCipher are:
- armeabi
- armeabi-v7a
- x86
To minimize the file size increase caused by SqlCipher you can build a seperate APK that only includes the architecture that matches your users device. Take a look at the Android developers site on how to publish multiple APK's targeting different device configurations.
During my testing following this technique reduced my APK from 10MB to 4MB.
For further information this article details how to implement SqlCipher on Android and explains the file size increase.
It depends on your actual needs but a simple way would be to encrypt/decrypt the data itself before you write to / read from the database. This would have minimal impact on the apk size but there's more effort needed on your side.
If high security is required, just accept the 10MB and get on with life :-)
There are two main databases are widely used currently for Java/Android applications. In this case, SqlLite and H2 Database. According to SQLite vs H2 DB we can see a set of differences features between both databases. For example, both databases has ACID features, transactions, referential integrity and Unicode representation. However, only the H2 Database has Encryption Data, Brute-Force Protection and Native Network Encryption.
This Encryption mechanism uses the AES algorithm where the database files can be encrypted. As H2 Database uses AES to emprypt data files, in general, we can infer that it can be lightly slow than SqlLite because en[de]crypt parses used by the AES. So, en[de]crypt can add an overhead cost in terms of efficiency when was used the H2 Database.
In summary, H2 Database can be an useful alternative to small projects when encryption data is important. In addition, it is important to notice that more experimental (empirical benchmark) is important to do conclusions about performance issues that involving booth databases. Notice that small and lightly slow are subjective since we need experimental evaluation to be more precise to compare both (or others) databases.
Some useful references used: