CODEWITHSUNDEEP
mongodbelasticsearchfull-text-searchbigdatadatabase
amin msh
amin msh

Reputation: 474

How can I handle a very large database and do not miss the performance?

if i want to develop an application, I'm worried about its performance after the number of users and stored data increases. actually I don't know what is the best way to implement a program that it works with a really large data and do some things like search in it, find and receive user information, search text and so on in real time without any delay !

Let's me explain the problem more

for example i have chosen 'Mongodb' as a database and suppose we have at least five million users and a user want to log in into the system, the user has sent the username and password
The first thing that we should do is to find the user with that username and then check the password, in mongodb we should use something like 'find' method to get the user's information, something like below:

Users.find({ username: entered_username })

then get the user information and we check the password but the 'find' method should search the username between million users and it's a large number and if any person request for authentication, this method should be run for each of them and it cause a heavy processing on the system

but unfortunately this problem is only for something like finding a user, if we decide to search a text when we have a lot of texts and posts on the database the problem is more bigger

i don't know how big companies like facebook and linkedin search through millions of data in such a short span of time. actually i don't want to create something like facebook or more but i have a large amount of data and i'm looking for a good way to handle it

is there any framework or something else that help me to handle large data on the databases or is there exist a method to implement data on database so that we search and find data fast and quickly? should i use a particular data structure?

i founded an opensource project elasticsearch that it help us to search faster but i don't know if i found something with elastic how can i find it on mongodb too for doing something like updating data and if i use elastic search i should use mongodb too or not!? can i use elastic as a database and as a search engine simultaneous !?
if i use elasticsearch and mongodb together then i should have two copies of my data, one in mongodb and one in elasticsearch!? and this two copies of the data that are separated :( i wish elasticsearch search in the mongodb that does not have to create two copies of the data

thank you if you help me to find out a good way and understand what should i do.

Upvotes: 3

Views: 2094

Answers (1)

kevinadi
kevinadi

Reputation: 13795

When you talk about performance, it usually boils down to three things:

  • Your design
  • Your definition of "quick", and
  • How much you're willing to pay

Your design

MongoDB is great if you want to iterate on your data model, can scale horizontally, and very quick if used properly. Elasticsearch on the other hand, is not a database. However, it is very quick for searching. A traditional relational database will be useful if you know exactly how your data looks like, and don't expect it to change much, or is relational by nature.

You can, for example, use a relational database for user login, use MongoDB for everything else, and use Elastic for textual, searchable data. There is no rule that tells you to keep everything within a single database.

Make sure you understand indexing, and know how to utilize it to its fullest potential. The fastest hardware will not help you if you don't design your database properly.

Conclusion: use any tool you need, combine if necessary, but understand their strengths and weaknesses.

Your definition of "quick"

How "quick" is quick enough for your application? Is 100ms quick enough? Is 10ms quick enough? Remember that more performance you ask of the machine, more expensive it will be. You can get more performance with a better design, but design can only go so far.

Usually this boils down to what is acceptable for you and your client. Not every application needs a sub-10ms response time. There's plenty of applications that can tolerate queries that return in seconds.

Conclusion: determine what is acceptable, and design accordingly.

How much you're willing to pay

Of course, it all depends on how much you're willing to pay for all the hardware that need to host all that stuff. MongoDB might be open source, but you need some place to host it. Also, you cannot expect magic. You can't throw thousands of queries and updates per second, and expect it to be blazing fast when you only give it 1 GB of RAM.

Conclusion: never under-provision to save money if you want your application to be successful.

Upvotes: 3

Related Questions