I'm working with a client and after we updated the KVM via a management API call there was a delay in the proxy's ability to retrieve the updated value from the KVM. The delay varied between users. My delay was about 5 minutes, but other users stated that they experienced a longer delay. This is a single data center installation.
I have couple of questions regarding this topic:
1) Is there documentation that describes the caching used by the KVM? Does the KVM follow a similar caching strategy as used by the Cache policies?
Documentation for the Cache (Ehcache) implementation is listed in the following link
http://docs.apigee.com/api-services/content/cache-internals
I read several posts that hint at the KVM caching strategy, so it would be nice to have more details on it.
https://community.apigee.com/questions/7251/inconsistent-results-from-kvm.html
https://community.apigee.com/questions/7800/kvm-suitability-for-storing-small-data.html
The docs also state that there is cache:
<ExpiryTimeInSecs> element
For a GET from the KeyValueMap, specifies the duration in seconds after which Edge refreshes its cached value from the KeyValueMap.
A value of -1 means the GET value never expires and, because the value never expires, Edge never refreshes its cached value from the KeyValueMap.
2) The following article mentions that every time a KVM lookup request is issued, then all the entries are retrieved, even though you specify the key. Has this been resolved or changed in later releases?
https://community.apigee.com/articles/1396/apigee-developer-heres-the-stuff-you-should-know.html
3) Is it possible to explicitly invalidate the KVM, and force a cache update, similar to the invalidate cache policy? If not, then is there a way to change how quickly this update propagates using the Apigee conf files?
I know that updating an entry in the KVM invalidates the cache eventually, but I'm curious to understand how this process works and whether or not it can be configured.
LinkedIn
Twitter
