The Next Big Thing

You may have noticed that the podcast has gone a bit silent over the last few weeks. There's a reason for that, I promise. Brian and I have been wondering, off-air, why we keep finding ourselves recommending "rental" more and more. It's all become a bit of rote - so we're shifting focus and re-targeting.

Hey. James here.

You may have noticed that the podcast has gone a bit silent over the last few weeks. There’s a reason for that, I promise. Brian and I have been wondering, off-air, why we keep finding ourselves recommending “rental” more and more. It’s all become a bit of rote – always rent, build your core competencies, and buy for scale.

While we could continue hammering on that formula, evaluating more and more of the cloud-native space from a functionality perspective, we thought we’d mix it up a bit. We’re going to dive more deeply into each of the large and large-ish cloud providers, and see what they provide if you go “all in”

So we’re going to talk AWS. GCP. Azure. Linode. Digital Ocean. In this “service-first” format, we’re looking to explain what these things that the cloud providers have actually do. What’s under the hood (mostly speculation; informed speculation)? How would you build a replacement? Why (and when!) would you ever want to do so?

As consultants to the larger enterprises, we’ve seen the promise of multi-cloud puff up and then deflate. As our (very large) clients went to the cloud, they wanted to control for the variation in cost, and tried to maintain a level of independence from any single vendor. Amazon threatening to raise EC2 rates? Shift those virtual workloads over to Azure! Google blob storage not resilient enough for you? Migrate those bits and bytes over to S3!

What few counted on was the sheer cost of the complexity involved in maintaining that independence. To truly remain “cloud-agnostic”, you have to purposefully cut yourself off from some of the “deeper” services that one provider has that other lack. Otherwise, you run the risk of depending on something that you are not prepared to migrate from vendor to vendor, not without a ton of late-stage infrastructural implementation work.

What you end up with is a lowest common denominator model of cloud services: disks, file buckets, virtual machines. You get to (have to) supply all the rest. Need an event bus? Here’s RabbitMQ; knock yourself out. Need a database? PostgreSQL is containerized, pop that onto a VM and then design all of the HA and DR/BCP stuff yourself.

In the end, you’re not really doing cloud; you’re doing virtual data center, and you’re hiring Terraform SMEs to glue and tape it all together.

At the other end of the market, SMBs and startups, we saw something entirely different. People realizing that shielding themselves from the vendors pricing whims was not a business-critical activity. “We just use Amazon” or “We are a Google shop”. Much like choosing phone service, or hiring a cleaning company. A commodity.

And then we realized, we’d been going about this podcast from the point-of-view of a cloud-agnostic firm that wanted hybrid- or multi-cloud, had the time to evaluate all of the alternatives, and make the right decision. That’s not even 75% of you!

So we’re taking a break, re-orienting, and focusing our next episode on what the providers provide. If all goes according to plan, we’ll drop episode 11 next month, on September 2nd.

See you then.
♥ James