We have services running on AWS. we are planning to expose them as APIs. I read the documentation that APIGEE Edge has greater benefits if both back end service and Edge running on AWS. My question is , Whether we need to use APIGEE Cloud version or APIGEE on-premise version (private cloud)?
Which one will have more benefits in terms of cost and performance ?
Whether APIGEE cloud is partnered with AWS for cloud hosting or APIGEE has its own cloud platform other than AWS ?
If you haven't already, I recommend take a look at our eBook, Apigee Edge: Apigee Cloud vs. Private Cloud that goes into some depth comparing the options.
And also I recommend get in touch via our contact form where we can give you a more tailored response based on your specific use case.
If I've answered your question, please click the Accept link or let us know how we can help further on this question.
Thanks for the clarification. There isn't any feature advantage that you would get, but there might be a potential difference you might see in latency between Apigee and your Backend/ESB services depending on which cloud platform you select. It is a bit difficult to be definitive because it also depends on which geographical regions you are looking at. Regions where GCP, Azure, AWS, etc have presence that's physically closer together will minimize latency differences for example.
If I may be so bold (and I imagine you are already thinking this way I just wanted to get it here too) is to evaluate the different Cloud Platforms on their own merits, find the one that best fits with your specific Backend/ESB use case, current and future technology needs, etc.
The variation in latency between the cloud platforms probably isn't going to materially affect most use-cases, but actually the best way to be sure of that is via a direct evaluation. Most cloud platforms have free tiers so you could set up a simple eval in each and look at the latencies between them - that would give you some indication.
And my last point on latency is that there are also ways that it can be mitigated, e.g. via caching so just to reinforce the point that latency by itself should not be a decider in most cases.