When designing and building your database overtime the tables sizes will increase and if over one million rows per table it can often really slow your application. First and foremost is building a relational database structure where foreign keys relate to one other tables and data is organized in tuples. The logic connections between these tables are organized quite similar to that of the animal kingdom. You can animals that are broken into species, types and sizes for instances each attribute is then is own table. Rather than trying to store all the data in one table try to break it out so its logical this will not only speed up your database but eventually create less code for the application when retrieving, reading, writing data, and keep the application simple to understand and scalable.  But still you may have more than one key to reference other tables for instance you are calling up the attributes of a lion this pulls from the species, type and size table so in this case you want to ensure you create indexes of all other joined data. Indexes should always be present for any other data that will be typically associated with queries that could be pulled from those tables. This is used by the database search engine to speed up queries rather than starting from the first row for your join and having to search the entire database to find the associated row sequentially you can quickly determine the position. You should always add your indexes before you database grows, as adding them after will take a very long time so do it before that is present. This will eliminate other rows that should not be considered and also use integer types for creating your indexs not varchar or other types.