Lately, I started working on a hobby project (Site Relic) to test website performance, taking a screenshot, security check from multiple locations.
Having server requirement in multiple locations means I have to consider the internal latency as a server will talk to each other to deliver the requested content from requested location.
Google Cloud Platform (GCP) is known for low latency, and I am happy with Geek Flare performance which is hosted on GCP.
You can check the latency to GCP over the Internet by using online latency tool such as Cloud Harmony.
However, I couldn’t find any resource which gives an indication of latency between Google Cloud regions.
So I thought to do a little test myself and here is the result.
Google Cloud is available in the following thirteen regions.
- Iowa, US Central
- Oregon, US West
- North Virginia, US East
- South Carolina, US East
- Belgium, Europe
- London, Europe
- Frankfurt, Germany
- Singapore, South East Asia
- Taiwan, East Asia
- Tokyo, North Asia
- Sydney, Australia
- Sao Paulo, Brazil
- Mumbai, India
I provisioned f1-micro instance in all the region with Ubuntu 16 LTS.
I didn’t install any additional software and did a ping to another region server 10 times over private IP.
ping –c 10 $PRIVATE_IP_ADDRESS_OF_OTHER_REGION
And here you go with the average latency results in milliseconds.
Google Cloud Latency Between Region
View the latency data in full page
- Green = Less than 100 ms
- Orange = Between 100 to 200 ms
- Red = Above 300 ms
- Sydney & Singapore got high latency from/to all the regions.
- Oregon perform better to connect to Asia
- Tokyo got the lowest latency to connect to Asia and US
- London play better in Europe
Above should give you an idea where to host your services internally for better results.