Apigee Microgateway - Capacity planning- Logs

Hi,

As per the link Apigee Microgateway Installation.

1.Does not list the minimum Disk required for Microgateway.

2.Do we have any minimum Hardware requirements to follow ?? and derive based on performance testing or any high level VM requirements to start with ?

3. Given any design how i ensure that, if I need to dockerize Apigee Microgateway or still go without Docker. Dcokerize dos and Donts gives listed details..

Do we have any other link from Apigee side as well based on usage of docker or containerizing Apigee Microgateway ? . Looks like having Containerizing may have the Performance issues..

0 1 172
1 REPLY 1

1. It is expected that you handle log file rotation on your data center yourself. By default, logs are stored under the /var/tmp directory but you can customize this in the yaml config under the 'edgemicro/logging' stanza. The best practice would be to use the linux logrotate deamon [https://linux.die.net/man/8/logrotate] to periodically compress and move files to a more permanent storage. Since Edge Microgateway doesn't store a lot of files other than the logs, the system requirements for disk space are very low. Try running load tests and then using the `du -h /var/tmp` command to see how fast disk usage grows.

2. It depends on the load and availability requirements. I would recommend you run it on at least two data centers or availability zones in production and use a DNS based load balancer in front of them. In terms of VM sizing it is flexible. It can be run on VMs as small as 1 core / 256 MB of RAM up to 8 cores / 32GB of RAM. Be sure to set the Nodejs heap size to utilize it all! (for example --max-old-space-size=8192). MGW will utilize all the cores available and spin a process for each one, although you can also configure that via the -p option. You can also use the Chrome inspect tool to debug and profile your application. To run mgw in inspect mode use:

node --inspect $EDGEMICRO_HOME/cli/edgemicro start -o <org> -e <env> -k <key> -s <secret> -p 1 

3. There are multiple benefits of using docker to run MGW. My favorite one is the ability to spin-up identical containers which enables you to scale nodes elastically to handle the demand. Containerizing will not impact performance. Others may have different opinions on this and there are plenty of discussions on docker on the web for you to peruse.