1

I just want to know, can Arangodb handle 20 billions of records in a single collection?

What is the best way to insert 1 billion records at a time?

Yash Pandya
  • 13
  • 4

1 Answers1

1

Of course. The underlying storage engine choice should definitely be RocksDB. How big would every individual document in average be?

The best way would be any sort of import/restore. Is this going to be a cluster or a single server deployment?

Kaveh Vahedipour
  • 3,412
  • 1
  • 14
  • 22
  • The individual document can be of size 1 kb. Currently I am using it as a single server and mmfiles as a storage engine. – Yash Pandya Nov 20 '18 at 13:40
  • MMFiles is not an option unless you are looking into buying 20 exabyte of RAM. Memory mapped files performance goes towards zero, when data size including indexes cannot be matched in RAM size. – Kaveh Vahedipour Nov 20 '18 at 17:45
  • Thanks for the reply. You suggested the best way to insert records is import/restore, but to use that my whole data should be ready before insertion,right? Suppose my data is generated at run time using some sort of script, at that time what should I use? – Yash Pandya Nov 22 '18 at 14:40
  • Absolutely. By all means insert all you data. – Kaveh Vahedipour Nov 22 '18 at 14:48