DX Unified Infrastructure Management

 View Only
  • 1.  Oracle Probe - How many profiles/checkpoints per probe

    Posted Apr 02, 2020 11:31 AM
    Hi guys.

    How many profiles and/or checkpoints I can configure per probe (oracle)?

    I'm planning to monitor some oracle databases, but I cant find any official documentation that show any information like that. Any sheet or formula to calculate?

    Thank you.


  • 2.  RE: Oracle Probe - How many profiles/checkpoints per probe

    Broadcom Employee
    Posted Apr 02, 2020 01:30 PM
    Nor have I, I don't think there is a "hard" limit in the probe or like you said not documented. I would not go crazy though and deactivate the ones you don't need.


  • 3.  RE: Oracle Probe - How many profiles/checkpoints per probe

    Posted Apr 02, 2020 05:13 PM
    Hi Gregg, thanks for your answer.

    Think about this situation: I have "X" databases and "Y" checkpoints to config and I have to send a request with the quantity of servers that will work like polling robots, because I have to buy these servers, so I don't know how to calculate how many polling robots I need. That's the point.


  • 4.  RE: Oracle Probe - How many profiles/checkpoints per probe

    Broadcom Employee
    Posted Apr 02, 2020 05:35 PM
    In the VM age I would say start small and add as needed. I have run about 150 profiles on 1 oracle probe with no noticable issues. The poller also had url response but no other big users of resources like netapp or vmware


  • 5.  RE: Oracle Probe - How many profiles/checkpoints per probe
    Best Answer

    Broadcom Employee
    Posted Apr 02, 2020 06:07 PM
    As of the latest GA version, the oracle probe has been reported to work well (as per some field services resources), with up to 400 profiles, all default checkpoints enabled, and up to 10 connections. Other reports from the field state up to 30 connections and a few hundred profiles (non-MCS). That said, there is no official/formal oracle performance and scalability benchmark.

    You could deploy the oracle probe on a hub or a robot but if there are several other 'busy' probes on the machine you might opt for a hub. I would suggest that you start by configuring a custom oracle monitoring template that only includes the key performance indicators (KPIs) and/or key metrics (checkpoints) that are a MUST-have to monitor. Then add a decent number of connections, e.g., 5-10 and associated profiles and see how the probe behaves at a specific monitoring interval, e.g., every 5 minutes. Enable QOS data collection and/or alarms where it makes the most sense. On a virtual server, you may be starting with 2-4 virtual processors, and 8 GB of memory. Assuming the server has 4 processors, and 8 GB RAM for starters, you can keep track of CPU and memory utilization and add more resources as needed. But keep in mind that its possible that you will reach some inherent limit with the probe regarding the number of connections and active profiles. The signs may be somewhat varied, but you may see profiles failing to execute or other signs that the probe has become unstable, such as the probe 'not running,' profile status icons turning yellow or black, or probe returning unexpected results. If and when it gets to that point, you can consider splitting up the load (profiles) across two or more hubs/robots.

    ------------------------------
    Support Engineer
    Broadcom
    US
    ------------------------------