1 of 1 people found this helpful
Can anyone share the challenges during deployment of ADDM in Cloud environment?
>>> Cloud resources can be scanned like on-premise hosts. So I guess challenges won’t be different from discovering on-premise host.
If you take example of the EC2, if all its instances are visible to the BMC Discovery appliance, and the required access is in place like ssh, snmp, windows access etc.. then they can be discovered. So challenges may come in terms of firewall, accessibility of instances to BMC Discovery.
If you are thinking of discovery of detailed integration with cloud providers, such as AWS, Azure etc. are not available at this moment. And based on few community posts it seems work is in progress.
There was survey conducted some time back.
2 of 2 people found this helpful
Whilst the Cloud servers can be discovered like regular on network servers the one additional challenge we have come across is how to deal with VMs where the Cloud systems allow the users to effectively put their VMs into a 'stalled' state.
ie A developer request a Cloud Server, its provisioned within 30 mins, we scan it, he uses it for a few days for a project, then to prevent being charged the daily usage fee, he is allowed to request the VM to either be decommissioned or to be put on a 'stalled' state for a set period of a few weeks etc and then some time later he comes back and request the VM to be put back to available, thereby preventing the need to reinstall various software components that are outside of the base images available on the cloud offering.
We are thinking about an appropriate way to get round this and not cause Discovery to get into a mess. At the moment the thought would be to have a prestep prior to scan to query the Cloud mgmt. systems api and check the state on the VMs and then auto inject the valid ones into Discovery for scanning, rather than scanning all a seeing them trigger aging and removal as they are still assets
But this is a new area for us, so we are also starting to look through the community posts as well to see what others have done.
We have our appliances outside the Cloud currently and are discovering the VMs that are being provisioned in the Cloud defined subnets.
Yes, You are correct.
I'm going to assume that you mean in a cloud then. I don't think it's an officially supported deployment scenario but I imagine the challenges are mostly the same, just reversed.
I can speak from the limited experience I've had when experimenting with deploying into a cloud (AWS). Deployed for use from the internet side with no direct route into our network. I used the AWS management portal for vCenter and deploying. All very simple and straightforward to create an AMI and deploy from that.
With the correct routes into the network from there (and not into it from outside of your network) then the challenges will be similar to that of a normal deployment and what David Farthing and Amrendra Singh raised.
1 of 1 people found this helpful
Agree with David Farthing on this one. In regards to discovering into the cloud, in this case AWS, it is technically feasible given a static EC2 instance that remains online and active, but this, at least in our environment, is not how cloud is being leveraged. Rather instances are spun up and run ad-hoc - think batch jobs, increased compute, etc, sometimes for a few days, sometimes less. We see multiple challenges here (ITSM too, not just discovery), especially around software licensing (time based, cloud friendly, call it what you will) and stale hosts due to system wide aging settings. At the very least, like the vmotion events integration, AWS integration would benefit from some kind of event trigger given an EC2 state change.. something that would allow discovery to keep up with the rate of change in cloud deployments.