-
Hello jithin,
First of all, Bandwidth should not be confused for throughput. There are a lot of factors that affects throughput such as latency, throttling of your ISP, packet loss, communication medium , and etc. It is really impossible to determine what your throughput will be as it is a case to case basis. But for the sake of the question, let’s pretend that we are in a ideal evironment.
Even with a great network connection, you can’t fully utilize its bandwidth. The limitations of the TCP/IP protocol make it very difficult for a single application to saturate a network connection.
So, in order to maximize the 30/MB bandwidth, we have to use multithreading which s3 multipart upload uses. We can break the 200GB file size into 10,000 parts each having a size of 20MB. Let’s say that we are using the default concurrency configuration of multipart upload which is 10 concurrent requests. Considering an ideal environment, each processes will have a throughput of about 24Mbit/s or 3 MB/s. A single 20 MB file would take 6.6666 seconds to upload. But since we are uploading concurrently, 10s of these 20 MB files are uploaded at a single time.
10,000/ 10 = 1,000 batches.
Each batch would take 6.6666 seconds to complete. 1,000 * 6.6666 is approximately 6,6666 seconds or 1.83 hours.
1.83 hours < 2 hours. So, yes. S3 multi-part upload is the correct answer.
Regards,
Carlo@TutorialsDojo
