Today, the Datanauts don our firesuits and see if we can brave the flames of do-it-yourself OpenStack. From what we hear, it’s a tough go, at least in the experience of some. I mean, maybe rolling your own OpenStack wasn’t so bad for you. But it was for our guest.
That guest is Tony Bourke, who was on a recent show talking about the fundamentals of cloud-native applications. He’s back to share his experiences rolling his own implementation of OpenStack. Our discussion covers the benefits of OpenStack and why the organization chose it, and then gets into challenges he faced in deployment and operation, including installation, upgrades, HA, and more.
This episode of Datanauts is brought to you by ITProTV. Enhance your technology aptitude. ITProTV is the resource to keep your I.T. skills up to date, with engaging and informative video tutorials. For a free 7-day trial and 30% off the life of your account, go to itpro.tv/datanauts and use the code DATANAUTS30.
Part 1 – For Labs, OpenStack Is Simple
- DevStack option
- 3-node option
- Physical requirements
- Hypervisor-on-hypervisor concerns
- Networking concerns
- VMware Integrated OpenStack
Part 2 – Rolling Your Own OpenStack Is Hard
- Disambiguation: Rolling your own versus running your own OpenStack
- Creating your own OpenStack distribution is like creating your own Linux distribution: You would only do it if you really, really, really, really needed to. There’s no operational benefit to 99+% of the uses cases in rolling your own. And it would create a lot of work that, for most shops, would provide no advantage.
- So you can pick a distribution and run OpenStack
- Whether you pay for OpenStack distributions or use the free ones, OpenStack is free as in freedom, free as in beer, and free as in a puppy.
- When things go wrong, there’s a lot of moving parts
- The Google well is not very deep (how many results do you get when you put in the error message, like a good engineer)
- Keeping up to date can be a challenge. Again, a lot of moving parts and interdependencies
- Running a single-node OpenStack is pretty easy to get going. HA is a bit different. OpenStack uses several databases and queuing buses, and it’s not prescriptive about them. The OpenStack administrators (and distributions) are responsible for maintaining HA of say, a MySQL database. So if a problem arises, an administrator needs to be familiar with a wide variety of technologies
- What gets monitored gets managed. Keeping an eye on all the log files, availability of services, etc., is a challenge itself.
Part 3 – Rather Than Build Your Own OpenStack, Use Someone Else’s
- Cisco Metapod
- Bluebox Cloud
- More easily fits OpenStack into a given organization
- Don’t need to staff/train up on OpenStack (can be a long lead time)
- Installations are typically much, much quicker
- Consume OpenStack as a Service (PCaaS, OSaaS, etc.)
- Public cloud experience, but in your own data center