In this video, we introduce PowerFlex 4.0 and its new unified PowerFlex Manager, which integrates all management and operations into a single application interface. Learn how to deploy and configure the PowerFlex management platform (PFMP) on your Bring-Your-Own ESXi environment. We cover everything from setting up management virtual machines (MVMs) to configuring DNS servers, deploying Kubernetes, and installing PowerFlex services.
PowerFlex 4.0 introduces a new unified PowerFlex Manager, which incorporates all management and operations into a single application interface. From a high level dashboard overview of the system and its health to detailed configuration and operation of block storage services, file storage services, snapshot management and remote replication resource groups, along with the declarative templates that deploy them.
We also have unified event and alert monitoring and detailed views of all discovered resources in the system. Unified user and role management, compliance, OS and network management, license and security management, and software updates. PowerFlex Manager is the user facing side of the stack. Underneath it is the new distributed containerized PowerFlex management platform, or PFMP. PFMP can run on top of various kinds of infrastructure. In this demo, we'll see your Bring-Your-Own ESXi environment, but the steps shown here apply almost equally to deployment on single or multi-node PowerFlex Manager Controllers.
In all cases except CoResident, the PowerFlex management platform is installed across three management virtual machines or MVMs. Using a supplied machine image, the MVMs must each be assigned a distinct resolvable hostname. Their time must be sync using a common NTP server. Assign an IP on the PowerFlex management network VLAN, preferably one that can route to the data and other networks. Ensure that your DNS servers are configured in each MVM. A temporary installer VM is used to deploy the Kubernetes platform and all the PowerFlex services onto the MVMs. Now let's log into the installer VM.
The supplied installer VM image has Docker and all the necessary tools pre-installed. Create a working directory and upload a copy of the PFMP tarball. We have done that already here. The PFMP package contains all the code necessary to set up the Docker based bootstrap environment, which will in turn deploy the PowerFlex management platform to the 3 MVMs. Unpack the PFMP tarball and then change into the PFMP installer directory.
Before we run the installation scripts, we must first set up the configuration file. An example file is provided. Note the things to be included. First, we must specify the hostnames and IPs for each of the MVMs we created. Second, there are two reserved IP pools which should be left as is. These IP ranges are dynamically used by the Kubernetes cluster for all pods and communications internal to the Kubernetes cluster.
The next section is where you specify IPs specific to your own data center and the environment you wish to deploy. The management pool contains IPs that will be used for the Kubernetes clusters external facing services. These should come from or be routable to the PowerFlex management network in your data center.
If this management subnet or range is routable to your data networks and the out-of-band network where the node iDRAC and the switches will live, then you do not need to specify those other networks. However, if the PFMP cluster needs to communicate with networks that are not routable from one another, then they must all be provided here so that the Kubernetes cluster configures its services and also listens on these ranges. Finally, you must provide the primary PFMP hostname and IP.
This will be the primary ingress IP, and is where users both log in to PowerFlex manager and the IP endpoint, where the APIs are queried. Modify the template to match the environment you wish to deploy. We have already created and saved our template, so we'll just swap it into place here. Here we see the IPs and host names for our MVMs, along with two routable IP pools in our lab.
The data networks are routable from the management network, so we don't require an additional configuration for those. However, the iDRACs for our nodes are on a subnet not routable from the management network. And we specify that range here. So PFMP will be able to communicate with and configure the nodes. Last we have the fully qualified domain name and IP that will be the main access point to PowerFlex manager. Now we're ready to begin installation. Change from the config directory to the scripts directory. Ensure that IP forwarding is configured on the temporary installer VM. Then run the setup installer script.
This sets up the containerized installation environment. When this is finished, we run the install PFMP script. The script reads the configuration file and asks for the SSH username and credentials for all the management VMs. The install PFMP script will configure the MVMs, set up the base Kubernetes platform, and then install and configure the PowerFlex pods and services. Go grab a coffee.
This whole process may take a couple of hours. For the demo, we will accelerate immensely. There are a couple of ways to monitor progress. You can run a Docker logs command to follow the installers progress. Alternatively, you can tell the file to which all the installation logs are written. This step is optional, but some of us love to watch as things progress. The first phase is now complete and the base Kubernetes platform is installed. After this, the deployment proceeds to install and configure the services specific to PowerFlex Management and Operations.
When the deployment completes, we have now fully configured and deployed the PFMP M&O stack. You may power down and delete the installer VM if you wish. From here on out, software updates to the PowerFlex Management platform are done from within PowerFlex Manager itself. The user interface for PowerFlex Manager can now be accessed at the URL specified in the configuration file. In the next video, we will log in and configure PowerFlex Manager.
