Pure1 Unplugged: Pure Storage moves Pure1 to dark sides - Pure1 solution for "isolated" Pure1 custom
[ NOTE: machine translation with the help of DeepL translator without additional proofreading and spell checking ]
Pure1 is Pure Storage's cloud analysis, management and monitoring tool. Pure 1, like all software features (even new ones), is always free! Pure1 is also the basis for Pure Storage's proactive support. Also, as should be already known, Pure1 is a cloud platform hosted and patched by Pure Storage, so there is no additional administrative work for your company.
For "Pure newbies": Pure1 does not represent the primary main management of the Pure systems. The systems can also be managed individually via any web browser using web GUI or CLI/SSH.
However, despite the highest security measures, there are still companies today that are not allowed to use connections to public cloud platforms due to various corporate regulations. Pure Storage is at no time able to use the transmitted META/telemetry data to read or restore your company data in any way. However, some companies are still not allowed to connect to Pure1 Cloud.
The consequence of this prohibition is that it is not possible for the affected companies to use Pure1.
Until July 2019 of the release of: Pure1 Unplugged.
So what is Pure1 Unplugged? In a nutshell: they unceremoniously turn the Pure1 public cloud platform into their own private-on prem Pure1 cloud platform. So if you are a dark side (Pure Storage), you can still use Pure1 Unplugged to centrally monitor and analyze all Pure Storage systems. Pure1 Unplugged addresses companies that would not be able to use Pure1 Cloud due to various corporate guidelines or security requirements. If regular use is permitted, the Pure1 Cloud route should also be used. Currently all FlashArray and FlashBlade systems are supported. Also as mentioned above: new features are all 4 free!
The basis for Pure1 Unplugged is Elasticsearch as a search engine, a program library based on Apache Lucence for full-text search (Twitter and Wikipedia also use Lucence), Kibana, Helm, Docker components, and Kubernetes. Elasticsearch is written on Java and stores documents in a NoSQL format. Elasticsearch is the most widely used search server, along with Solr. Kibana, on the other hand, is a browser-based open-source analytics platform built on top of Elasticsearch. Previously collected Elasticsearch information (available from index in No-SQL format) can be searched and visualized and dashboards can be generated from it. "Unplugged" can also be accessed like the regular Pure1 via a web GUI. Data collection/metrics is done via the REST API of the Pure Storage systems. It is Pure1 Unplugged is a GitHub project.
However, there are a few limitations:
As you can see, the supported Pure1 Unplugged features are not yet equal to Pure1 Cloud ... but I suspect that more and more features will be added here in the near future.
In addition, I also suspect at this point that the on-prem Mediator/Quorum for ActiveCluster will be integrated into Pure1 Unplugged. To date, this must still be deployed as a VM/OVA dedicated at a third datacenter location for dark-side customers. The use of the on-prem mediator is also usually only used for such customer requirements. Therefore, an integration is very obvious.
Pure1 Unplugged Installation
Pure1 Unplugged is available for download/installation as ISO and OVA. I myself prefer the ISO installation, because you can already adjust settings for the network etc. here. With the OVA these must be changed manually in the operating system via the command line. The base of the software is a CentOS 7 (64-bit), so Pure1 Unplugged can be installed on any CentOS capable operating system.
Pure1 Unplugged can be downloaded by any logged in and authorized user via the Pure Storage support portal.
The system requirements vary depending on the number of systems included. Under 10 systems, the minimum requirements are as follows:
50GB hard disk space/vdisk
Pure Storage says that Pure1 Unplugged's system programs will not work with fewer resources ... in the course of my tests, however, the system also ran within my VMware workstation with 8GB RAM and could not find any problems. Decreased values, however, I can only recommend in test environments! As mentioned above Pure1 consists of 4 powerful applications, which of course need appropriate resources for the performant work.
From 10 systems upwards at least 100GB disk space/vdisk should be allocated. Officially supported platforms are VMware Fusion and VMware ESXi (see below for more information - Hyper-V might also be officially supported soon).
For Purity versions Pure Storage specifies:
for FlashArray's min. Purity//FA 4.8.0 with REST API 1.7
for FlashBlade's min. Purity//FB 2.2.9 with REST API 1.5
Basic installation via ISO
I created a VM with respect (almost ツ) to the system requirements. I mounted the Pure1 Unplugged ISO and booted the VM "as normal".
There should be no problems here, the boot process succeeds without further doing if the option (connect at power-on - of the virtual CD-ROM drive in VMware) is set.
The installation GUI is loaded from the ISO, the installation language is selected, and the keyboard layout changes automatically in the background.
Further settings can now be made via the menu. First we change the time settings - the NTP settings can only be made after setting up the network adapter (IP address) (see note screenshot):
The mounted ISO is automatically used as the installation medium - no settings need to be made here:
Further software packages for various platforms can be installed via the "Software selection" menu. (At this point it can be seen that possibly even more platforms are supported as operating systems - HyperV, ...). ). I installed here directly the VMware Guest Tools with.
We select the previously created vmdk as the installation target. Kdump as a crash dump tool for system problems is active by default.
Finally, we adjust the network connection. This must be activated via the slider and should be configured with a static IP address. The hostname (as FQDN) I have also adjusted in this step.
In the menu at the bottom-right the button "Start installation" will change from gray to blue and you can start the setup. During the setup, the root access must now be provided with a password (for ISO, for OVA root with the password "password" is default). You can also create additional users.
Finally, the VM must be restarted once, which completes the basic installation.
Basic installation via OVA
In contrast to the ISO installation, a 100GB vDisk is already provisioned with the OVA. The installation and the keyboard layout are set to English and the root user is deployed with the default password "password". A DHCP address is obtained and you have to configure the IP address manually via CentOS to static. The VMware tools are also installed automatically.
The rest is "totally" unspectacular: Import OVA, confirm the default settings - actually next-next-finish - provision more resources if necessary (CPU, RAM) and then start the VM.
Pure1 Unplugged Setup
After the basic installation was successfully completed, you log in to the Pure1 Unplugged command line via console or SSH (I had done all further steps directly via the remote console).
For vDisks larger than the default (ISO-50GB, OVA-100GB), the root volume must be increased. This is done with the command "lvextend -r pure1-unplugged/root /dev/sda2" and finally checks the volume size with "lsblk".
We now have to edit a configuration file without further ado. I do this with the vi editor (nano would also work). I do this with the command "vi /etc/pure1-unplugged/config.yaml":
Who is not familiar with the vi editor, with the key "i/I" we get into the INSERT mode (leave via ESC - back to command mode). By pressing the ":" key (input must be a DOUBLE POINT!) we can be in command mode and can edit/save the file.
In the opened editor we change (key "i/I" - INSERT-Mode) to the line "publicAddress" and enter here in the ideal case the future FQDN of Pure1 Unplugged (because I had no FQDN entry, I took the previously assigned static IP address). In addition, the commenting of "podCIDR" and "serviceCIDR" must be removed. These IP networks are used internally for Pure1/Kubernetes to run various container services (these must not be terminated).
Other settings can also be made, such as adjustments to e.g. retention times of metrics, alarm data, logs, audit intervals etc.. An external authentication service (LDAP via Dex) can also be set up.
We save our changes via DOUBLE POINT "wq" -> Save file and exit vi editor.
We then proceed with the installation of Pure1 Unplugged by typing "puctl infra init". Here, Kubernetes is now configured accordingly.
After initializing Kubernetes, the Pure1 Unplugged service can now be finally configured: "puctl pure1-unplugged install".
Via https:// IP address or FQDN we now reach the Pure1 Unplugged web interface.
There are two login options, "Log in with Email" is for production use and "Log in with Example" is more or less in test/demo mode without authentication. I log in with "admin@example .com" and the default password "password".
After logging in, you end up at the "naked" dashboard. Here we first add our systems "Add New Array" (here: FlashBlade).