I've recently done a refactor of a org's code base and in bulk testing I initially used the dataloader to insert records, however for testing updates I decided to try using the execute anonymous tool in the developer console - mostly for ease of just copying and pasting a script in.
However I was doing with the (incorrect) assumption that execute anonymous would split the list of records into batches of 200 (I was working with 1000+ files). After a small period of concern that the changes I had made had actually reduced efficiency I realised that it wasn't doing this when comparing the limit usage and doing the same testing using the dataloader.
Now my question, with the assumption that I am correct about execute anonymous above (feel free to correct me!), what approaches do people take to bulk test their code? Would you consider dataloader insert/update etc with a batch size of 200 as sufficient? Any other approaches/considerations?