X

Best Practices from Oracle Development's A‑Team

Bandwidth and FastConnect for Data Transfer: Quickly and Easily

Overview

The purpose of this blog is to provide an understanding of how to calculate the required bandwidth for data transfer and what considerations need to be made end-to-end in Oracle Cloud to facilitate this transfer.

This blog will assume that FastConnect will be the primary connectivity strategy as the internet and various gateways are unpredictable and inconsistent for measure and calculation requirements.

Bandwidth, Latency, and Time

Several factors come into play with bandwidth requirements, and it's not just the speed of the connection that is key.   You can start with the furthest destination component and work your way back to the transfer source.    For Oracle Cloud, starting and the Compute Shape, which may be the final destination of the transfer, is an excellent place to begin.    

Below is a comprehensive list of Oracle Compute Shapes, and paying attention to the Maximum Bandwidth is a good starting point.  

https://docs.oracle.com/en-us/iaas/Content/Compute/References/computeshapes.htm

The next element that we need to look at is the speed of FastConnect. Again, customers have several options to choose from, which can be either 1 Gbps, 10 Gbps, 100 Gbps, or even flexible numbers from service providers that can use a 1 Gbps port for FastConnect but allow only 400 Mbps of traffic over the circuit.    

All of the numbers we are looking at are in bits, not bytes.   It's effortless to get the two of these confused, but we will stick to bits for this blog when talking about the connection points.

Now let's take an example of how much bandwidth I might need for transferring 50 Gigabytes of data over a 1 Gbps FastConnect circuit going to a VM.Standard2.1 shape, which has 1 Gbps of bandwidth capacity.

Use the below calculator: 

http://wintelguy.com/wanperf.pl

Set Link Bandwidth to 1000 Mbit/s
Set File Size (MByte):  50000

The result with no latency is around 07:01.  The reality is with all transfers; latency will exist.   Ping can be used to extract the Round Trip Time (RTT) of a packet between source and destination.  This result can be used to understand the latency between source and destination.  The calculator above can determine the data transfer time with the latency factor if we used the same numbers above and added 80 ms of latency; it now takes about 45 minutes to move the same data set.

As with many of these tools, this is only considering a single data flow with a default MTU of 1500 utilizing the total bandwidth available.  It is certainly possible to adjust MTU's and factor in other flows regardless of one or one hundred, as all of these will impact how much total bandwidth is required.   

Requirements of time, latency, or other variables can control how much bandwidth is required but understanding tools and the basics help in building a baseline for needs.

Summary

The information in this blog should provide a baseline for the required amount of bandwidth for FastConnect.  Furthermore, bandwidth for FastConnect can be altered based on time and latency constraints to meet the proper business objectives.   Bandwidth for FastConnect can be increase or decreased by simply changing the port size.  So if you require a sizeable one-time transfer over FastConnect, start with a more significant connection and downgrade to steady-state requirements.

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha