We are working on our new vehicle tracking system. We have been making use of mySQL for the past two years. However we have been discussing moving to a noSQL database.
We have a high number of reads and writes. For reads we have been told that mongodb will take up to much of our available ram on our servers. So which noSQL will best suit our needs, without causing us any furthur problems? Or is it better to hang on to mySQL?
Thanks in advance!
The first question is to decide why you want to move to a noSql database. Moving just because its the new way of doing things may not be your best option. The first thing to decide is what a noSql database gives you that traditional Sql can't. I would suggest that for a vehicle tracking system the answer is not a lot.
I would probably stick with Sql but might look at other options such as Postgresql with the Postgis extensions if some of its functionality makes things easier.
What is the purpose of the application? To give advice on what NoSQL database to use that is necessary information. For instance:
1) Do you need ACID transactions?
2) What is the ratio between R/W?
3) Throughput (how many R/W per sec)?
4) Number of simultaneous users?
5) Environment (Java/Microsoft)?
6) How do you plan to connect to the database?
There are some great options in the NoSQL/NewSQL market but all databases have their pros and their cons.
If you need a transactional database that has the same level of transaction security and ease of use, I would go for a memory centric object oriented database. Even if they are memory centric they often secure all transactions on disk to fulfill the ACID (D=durability). Depending on your environment there are some options.
You can take a look at Starcounter that gives you true ACID, works In-memory, have a very easy to use API, easy to connect to and have extreme performance.
If you need Linux and Java environment for your servers, there are other options!
