Identifying Rogue proxy to minimize the environment impact

PNayak
New Member

Hi All,

Within our organization, we have multiple API proxy build in our Apigee ( version 1.6.9) cloud environment which are for different divisions. All API Proxy share the same cloud space.

We are trying to avoid a scenario where there is a Rogue proxy being build, or a proxy contains improper implementation, which can affect the other proxy or the cloud environment itself or performance of the environment.


Are there any manually checks we can perform prior deploying the proxy to PROD so that we can minimize (or avoid) the situation where the whole environment is unavailable? Are there any specific Antipatterns in proxy design we can check?

 

Thanks in advance,

Pari

0 1 52
1 REPLY 1

Even that you tried to specify your scenario, it's still wide open for interpretation on what needs to be checked. 

One of the things I can recommend, is the Apigee Lint project, https://github.com/apigee/apigeelint

Which does all kinds of static analysis, which fits into your requirement of ' checks we can perform prior deploying the proxy to PROD'

The tool currently contains dozens of Best Practices as well as Best Practice violations that it checks for. 

A useful feature is its extensibility mechanism, external plugins so that you can add euristicts you've developed for your company. 
Apigee Lint is CI/CD friendly and it is recommended to run it as a part of your ci/cd pipelines.

Two extra references:

 

There is a book of antipatterns and many of them are performance related:

https://cloud.google.com/apigee/docs/api-platform/antipatterns/intro

 

There is also a Sonar Qube Apigee plugin, developed by one of the Apigee customers, if you prefer this technology solution https://github.com/CreditMutuelArkea/sonar-apigee-plugin/releases.

And final consideration, if you have suspicion that some proxies might potentially misbehave, (ie, a noisy neighbour problem), as part of remediation steps, you can use X/Hybrid environment to isolate them and therefore minimize/control impact.