What is best mechanism to store libraries and load to bitbucket pipeline in Atlassian cloud ?

[yogesh_devi]

Hello,

We are moving our builds from on-premise Bamboo to Atlassian cloud (bitbucket pipeline) 

Our build involves linking  third party and locally build libraries

For on-premise builds (using Bamboo) we have been  hosting these libraries in a local repository . bamboo build downloads the libraries for the specifc software product and version being build from the local repository.

Since we're moving to  Atlassian cloud (bitbucket pipeline), these libraries need to be available to be downloaded during builds running on bitbucket pipeline.

Our IT team does not encourage opening up our internal repository to internet owing to security reasons. They suggest that we take the needed libraries and host them elsewhere.

So What is the suitable option for hosting the needed libraries so they are available to build  happening on bitbucket pipeline in Atlassian cloud ?

Some option that I thought are listed below - but they come with some cons 

Option 1 :  use a virtual machine in public cloud to host libraries. Con - in addition to charge of VM running,  we will also incur cloud data egress charges when pipeline builds  download the libraries 

Option 2 :  I wonder if its a good idea to create version specific docker container images and pre load the libraries on that docker image. When use  container from that docker image for build it will have the libraries available, However the libraries total size runs in a few Gigabytes  ( up to 3 GB) , is it a good idea to have  3GB docker container ? Con: We will have to continue to update docker containers as we refresh libraries of make new release 

option 3 : STore libraries in a cloud bucket ( like a GCS bucket in google) and use them from there .   The problem is I havent found a suitable pipeline to copy code from GCP bucket to the build environment 

option 3 : STore libraries in a bitbucket itself.   Con. It isnt a very good practice to store large libraries in GIT version control .. but your opinion is welcome 

option 5 : Use local runners on premise.  I wonder if this will again need the linux runner host machine to have an internet facing IP , And moreover will a runner have access to my local IPs ?

Does atlassian provide a place to host stuff needed for builds that the pipeline  can download from ? 

[yogesh_devi · Accepted solution]

I placed the libraries in a GCS bucket - and just used a simple WGET to sopy thos libraries. So basically - as far as you can exercise right access control based on whether your content is ok to be publically accessed or you want to limit access to pipeline -- then placing it in a bucket is a workable solution