As the expression goes, “no rest for the weary.” A lot of us in the IT industry can relate to that statement. We are often needing to work long hours with heavy workloads.
Why is it then, that at least in my observation, many systems and network engineers still have not jumped on the automation train? We’re so busy being busy, that we don’t take the time to save time. Some of us still think RESTful APIs are something that only “devs” use. Luckily for our sanity, many of us are picking up this much needed skill.
If you haven’t played with RESTful APIs yet, and are in IT operations, I highly encourage you to check out RESTful APIs and get some lab time. By sharing some personal experience, I’d like to argue that there really can be “REST for the weary.”(Pun intended)
I remember before RESTful APIs were something that more vendors were supporting. I had to do write Expect scripts in Linux to do my automation. Really, this was just screen scraping terminal output and brute forcing automation. It was messy at best, especially trying to automate devices that weren’t made for automation.
This wasn’t a whole lot of fun, but it was worth the pain involved since it still saved mountains of time. There was quite a bit of pain with some vendors. They needed weird keystroke combinations before it would allow CLI access even after SSH’ing in. Implementing “CTRL+Y” through an SSH session via script was way more of a headache than you’d think it should be.
Complaining aside, spending time writing Expect scripts was certainly MUCH easier than doing things manually. As an example, in a large environment I previously worked in, my co-workers spent time manually SSH’ing to literally hundreds of switch stacks. They had to run some commands and capture output to save.
I was asked on a regular basis by management to get MAC address/switchport/dot1x info, and other data which could be queried for historical data in a database. This was in preparation for a forklift upgrade on the network. The use case was to get a history of devices, interface information, and all other relevant historical data. Pulling MACs from the cores via uplinks wouldn’t give necessary detail.
This wasn’t an option. We had to go to the switches as a source of truth. The idea was also to compare to data from other sources (the IP-PBX system was one example) in preparation for the upgrade. They wanted to make sure network cut-overs were un-noticed by end users, aside from the downtime to swap equipment of course. I spent quite a bit of time writing and tuning Expect scripts, but still much less than doing things the old fashioned way.
Fast forward to my next job managing literally thousands of server nodes in a high uptime environment. I started getting asked to do things like update BIOS settings ON EVERY SERVER. To make things worse, as software engineering changed their code, they’d ask me to change BIOS settings multiple times. There was no way I was going to iLO into every one of those servers, reboot, wait, press F9 to access system utilities, select BIOS/Platform configuration(RBSU)… etc. Thankfully, I didn’t have to draw a line in the sand and explain that I was unwilling to do this.
After some research, I learned that HPE makes a RESTful API available on their Gen9 and Gen10 servers. Lucky for me, we were using Gen9 servers at the time.
Managing BIOS settings isn’t the only thing you can do by the way. You could probably integrate this into your monitoring system if it allows, for non responsive devices as an example. We’ve all seen servers where the lights are on but nobody is home. They seem dead, but respond to pings…intermittently and with high latency. You could use your monitoring system to poll services, and metrics like CPU and RAM utilization. Then, reboot the server via iLO RESTful API, if it is really locked up. No more waiting for a human to notice alerts, escalate if needed, then reboot the server manually.
HPE Integrated Lights Out (iLO) server management provides intelligent remote control automation through scripting or an API. Gain even more capabilities that go beyond scripting by leveraging one API to manage your complete lifecycle of HPE Gen9 and Gen10 servers—iLO RESTful API.
A single API interface integrates server management components and full compute power. Use it with HPE iLO 4 and iLO 5 to perform remote server provisioning, configuration, inventory and monitoring to industry standards through Redfish API conformance.
This was a game changer. I was able to prepare some standard JSON files with standardized BIOS settings. Then, write a script utilizing HPE’s RESTful API, and push settings described in the JSON file to every server (or subset of servers for testing) with ease.
Some settings still required a reboot to take effect. This was easily handled by scheduled reboots during a maintenance window. I also didn’t need to patch together a solution to script iLO changes with the software equivalent of duct tape and bubble gum, but instead utilized a RESTful API. Something that would have taken FOREVER, was accomplished with ease using something very well documented by HPE.
Ready to give RESTful APIs a chance on your HPE servers and Aruba networking? Check out these resources: