Public cloud technologies can be used to improve our products and their operations even if they are deployed on-premise. Public clouds have become popular for their programmability which endear them to developers while providing unparalleled benefits such as no maintenance, lower total cost of ownership, global reachability and availability for all on-premise instances, unified and seamless operations monitoring via system center tools and improved performance via almost limitless scalability. 

The public cloud acts like a global sponge to collect data across on-premise deployments that serve to give insights into the operations of those deployments. Take the use case of collecting metrics such as total size and object count from any customer deployment. Metrics like these are lightweight and have traditionally been passed into special purpose on-premise storage stacks which again pose the challenge that most on-premise deployments are dark for monitoring remotely. On the other hand, the proposal  here is that if these choice metrics were to fly up to the public clouds like pixie dusts from each of the customer deployments, then they can improve visibility into the operations of all customer deployments while expanding the horizon of monitoring and call home services. A service hosted on the public cloud becomes better suited to assuming the role of a network operations center to receive callback-home alerts, messages, events and metrics. This is made possible with the rich development framework available in the public cloud which allows rapid development of capabilities in the area of operations monitoring, product support and remote assistance to customers for all deployments of software products and appliances. It also lower costs for these activities. Metrics is just one example of this initiative to move such developmental work to the public cloud. All services pertaining to the operations of customer deployments of our products are candidates for being built and operated as tenants of the public cloud.  

The proposals of this document include the following: 1) improving visibility of on-premise appliance deployments to services in the public cloud, 2) Using the public cloud features to improve manageability of all deployed instances, 3) Using a consolidated reporting framework via services built in the cloud that map all the deployed instances for their drill down analytics and 4) using the public cloud as a traffic aggregator for all capabilities added to the on-premise storage now and in the future. 

Specifically, these points were explained in the context of knowing the size and count of objects from each deployed instance. Previously, there was no way of pulling this information from customer deployments by default. Instead the suggestion was that we leverage the on-premise instance to publish these metrics to a service hosted in the public cloud. This service seemed to be available already in both the major public clouds. The presentation demonstrated Application Programming Interface aka API calls made against each public cloud to put and get the total size and count of objects. 

These API calls were then shown as introduction to several powerful features available from the public cloud for value additions to the manageability of deployed instances. The latter part of the presentation focused on the involvement of public cloud for the monitoring and operations of geographically distributed on-premise customer deployments.  

As a mention of a new capabilities made possible with the integration of on-premise deployment with manageability using public cloud, it was shown that when this information is collected via the API calls made, they become easy to report with drill downs on groups of deployments or aggregation of individual deployments. 

Finally, the public cloud was suggested to be used as a traffic aggregator aka sponge for subscribing to any light-weight data not just the size and count described in this presentation and thus become useful in alleviating publishing-subscribing semantics from on-premise deployments. Together, these proposals are a few of the integration possibilities of on-premise deployments with capabilities hosted in the public cloud.

#codingexercise

Determine the max size of a resource from allresources


Int getMaxResourceLength(List<String > input) {
return input.stream().map(x -> getSize(x))
                          .max(Comparator.comparing(Integer::valueOf))
                          .get();

}