I am conducting a study about evaluating Document database (MongoDb) and Relational database. For both databases, I created a simple application in web2py with only CRUD functions to evaluate fairly. However, I cannot find a good way in evaluating the two. I am mostly focused in terms of their Data Integrity, Ease of Use, Cost (in-memory) and Database Structure.
I need to find a reference on how to measure the 4 criteria. For instance, how can I measure the data integrity of Document database?
There are a lot of criteria to do so but to the best of my knowledge I don't know a benchmark you can use for this case.
But you can at least choose a specific domain and start by migrating the data from relational database to document database (mongoDB in your case). Once you migrate the data, you will need also to translate the main queries in the relational system and evaluate the following:
1. The execution time of the queries in both the relational database and the document database. You'll need to address queries of different complexities and with different resulting structures that can identify the difference between both models.
2. The accuracy/effectiveness of the queries.. making sure that they give you the correct and complete answers in both cases.
3. Size of both databases.
4. Consistency of both databases and if consistency is a main requirement in that domain, then obviously you will not be able to use MongoDB because of update anomalies that might arise after any update query.
I can think of a lot of criteria that you can use in the comparison but they are all dependent on the domain that you will be using. I'm not sure if you can use it independent of the domain. You can for example think of replication and using different shards inside MongoDB versus horizontal and scalable fragments replicas in the case of relational databases.
First of all, thank you for your responses. I really appreciate it.
To answer the question/clarification of Mr. Stefan, I use Relational Database and Document (MongoDB) for profiling database. I used Web2py framework and added only the CRUD functions. In my case, the application is a Profiling Application where I can Create a profile of a person, View their profile, Edit and Delete. I will be using the same application for both databases. I am having problems on how I can measure Data Integrity, Ease of Use, Cost (in-memory) and Database Structure in both databases. I need a checklist or evaluation sheet or steps in say, measuring data integrity, something like that. Please. :)
You can find different benchmarks that used different criteria to evaluate the performance of a relational database. The following are some examples of such benchmarks:
Relational DB benchmark: http://www.vldb.org/pvldb/vol7/p277-difallah.pdf
Relational Vs NoSQL: http://www.epcc.ed.ac.uk/sites/default/files/Dissertations/2012-2013/RDBMS%20vs%20NoSQL%20-%20Performance%20and%20Scaling%20Comparison.pdf
I hope this helps :)
It's also good to work into the direction of the TPC (http://www.tpc.org/). It's widely know, acceptable benchmarking practice with very robust parts.
In the paper recommend by the Neamat El-Tazi "OLTP-Bench: An Extensible Testbed for Benchmarking Relational Databases" they are also considering part of the TPC called TPC-C.
P.S. Many thanks to the Neamat El-Tazi for recommending "OLTP-Bench: An Extensible Testbed for Benchmarking Relational Databases".
You can read a book titled the fundamental of a database design for prof Ramez el masri.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs