Review the current market for functions and learn about the common risks associated with serverless architectures.
- [Instructor] So the current functions market is really active, and this is good for us as consumers because all the major Cloud vendors realize what's at stake here. Whoever gets selected as a leader in this category is going to gain the lion's share of the serverless business. Of course AWS is the market leader because they launched the category back in 2014. And they are building new features rapidly around AWS Lambda.
And we'll be exploring how some of those new features will impact your serverless architectures as we drill into the services in the next section of this course. Now in the world of Microsoft Azure, as they traditionally have done, they're building in addition to their core functions higher level functionality such as logic apps, templates, reference architectures, and tools very very quickly. This is increasing the adoption because of course it's addressing the needs of training and understanding how to properly implement these services.
Newer on the scene is Alibaba. They have released function compute as of this year and they're building template examples so that their customers can become productive with functions. In my work as an architect, I had one of my bioinformatics customers build out a serverless architecture on AWS and then extend that work and replicate that functionality pretty quickly on the Alibaba cloud for some collaboration with colleagues in China.
And to complete the picture, GCP has finally released their functions into GA. It was a long process. As we'll see when we drill into the services, they're focusing not only on functions but also at a lower level, on containers and container management with Kubernetes which came out of work that they originally did. A trend that I see around GCP is although they have core functions available they tend to build serverless services such as machine learning as a service or no SQL databases as a service.
They've tended to do that at a greater frequency than enhancing core function capabilities. Now that may change in the future, but that's the state of the market as I see it. So let's address some of the risks of serverless architectures head on so that when we work with architectures we can consider the implications. One of the ones that I hear really frequently is vendor lock-in. I think frankly it's overblown. The value that serverless services provide by reducing service costs and reducing management overhead is substantial.
Another risk that I hear about is immaturity of ecosystems, lack of tooling for monitoring, security, management. This one I think has validity. And we will look at this in detail around some of the services that we cover. Another risk is that my customers can't migrate current legacy monolithic applications. It would cost too much to move that data retrain people and rewrite the application. And that's very valid. That is actually a situation where we will either not use serverless architecture or come up with some sort of blended architecture.
Another one that's very subtle but that I've seen in production is as you break monoliths down you end up with a lot of small parts. And although if they are properly separated you should end up with faster deployments sometimes you don't because it's difficult to decouple and to create parts of a system that do not have dependencies on one another. And I want to close out this section with us thinking about this question of vendor lock-in.
What I'm seeing around the cost of functions and the cost of object storage is what I call a race to the bottom. The vendors are lowering the price around these two core services as much as possible because think about it. If you put your data in a particular vendor cloud you're less likely to move it to someone else's cloud. And if the vendor can then make money by charging for what I call vanity compute on top of that, so database segregation, machine learning, processing, so on and so forth.
That's a good business model. So the reason that some of the services are so incredibly cheap is the vendors are trying to incent you to put not only your compute but more importantly your data in their cloud. So we're going to talk a lot in addition to compute in this course about different options around data storage. And having the greater level of flexibility so that you can reduce your chances of lock-in is actually a very subtle but important part of serverless architectures.
- Defining serverless architecture
- Serverless services
- Serverless functions
- Cloud-native serverless architecture
- Serverless architecture for big data and machine learning
- Emergent serverless architecture