You are here

You are here

OpenStack remains essential: How to get your private cloud builds right

public://pictures/tytus.jpeg
Tytus Kurek OpenStack Product Manager , Canonical
 

When SUSE announced in October that it was pulling out of the OpenStack market, it renewed industry speculation that the once massively hyped technology is sliding toward irrelevance.

Nothing could be further from the truth.

OpenStack's alleged decline has become one of tech's favorite tropes. The infrastructure-as-a-service model of hosting private clouds was all the rage in the early and mid-2010s, touted (unfairly) by some as an alternative to public clouds.

That early excitement faded when OpenStack gained a reputation for being complex and labor-intensive, preventing many proof-of-concept projects from reaching production.

More recently, a newer kid on the block, Kubernetes, was often said to be overtaking OpenStack as the most vital cloud technology; this is a misperception. While SUSE cited a desire to "increase our focus on the growing importance of cloud-native and container technologies" and "align with technology trends around application delivery, Kubernetes and DevOps," you shouldn't read too much into what the organization's decision says about OpenStack's long-term prospects.

As a company that has implemented in excess of 100 private cloud builds, we've learned a lot about what companies should do to make their OpenStack deployments successful.

Here are seven important pieces of advice.

1. It's all about automation 

While OpenStack once was notorious for being complicated and labor-intensive, new automated features are changing the narrative. For example, if an enterprise needs to add more nodes to scale out, there is an OpenStack implementation based on charms that can do that without providing an extra layer of complexity.

2. Don't overcomplicate

Simple OpenStack architecture should combine OpenStack's control services with compute, storage, and networking services on the same nodes, rather than having specific nodes for each function. Machine containers allow you to do this effectively and flexibly, with the ability to easily move microservices around.

3. Go all in on LMA

OpenStack can be integrated with several third-party solutions for automated logging, monitoring, and alerting (LMA), patches, etc. This minimizes the number of people you need to do this work.

4. Don't build snowflakes

Teams using OpenStack shouldn't try to be heroes and build customized solutions, which can prove costly and difficult to operate. Make sure components and architecture are standard.

5. Create repeatable processes

Choose an OpenStack deployment that allows redeployment and reusability. For example, you should be able to redeploy OpenStack in any environment after having deployed it once.

6. Simplify hardware

In most cases with OpenStack, a single bill of materials for hardware is enough. Optimize hardware for cost efficiency.

7. Make sure your container strategy is multi-cloud

If you choose Kubernetes for container coordination, be sure to select a distribution that can work on top of an all-of-the-above list—​OpenStack, public cloud, VMware—​to assure multi-cloud portability. OpenStack's journey the last decade has been a bit of a roller coaster, but it is enjoying new vitality in the multi-cloud era. 

Emerging from the 'trough of disillusionment'

There's ample proof that, as it nears the tenth anniversary of its 2010 initial release, OpenStack has emerged from the "trough of disillusionment" phase of Gartner's hype cycle and continues to find sustained relevancy as open-source infrastructure in private clouds.

  • Many of the best-known companies on the planet across a range of industries rely on OpenStack. These include AT&T, Verizon, and China Mobile in telco; Walmart and Target in retail; and others such as Nike, Bloomberg, American Airlines, and Volkswagen AG.

  • According to recent statistics from the OpenStack Foundation, though OpenStack users often employ the three major public clouds, the average user devotes 58% of their infrastructure to OpenStack.

  • The OpenStack community—​one of the open-source world's most active—​has worked to address the complexity complaints that had dogged the technology. OpenStack's Day-2 operations story has never been better, and this maturation is a big reason for the technology's continuing traction in enterprises.

Multi-cloud takes a team

People should stop saying that OpenStack is losing to Kubernetes. These are not competing technologies, but complementary ones, especially in multi-cloud environments: OpenStack handles infrastructure, and Kubernetes is for containerized applications.

In fact, Kubernetes remains the top container-coordination platform running on OpenStack. It's not a zero-sum game.

Keep learning