Is PaaS serverless? It depends

6:10 · Watch on YouTube ↗

Transcript 799 words · about 5 min to read

Auto-generated captions from YouTube, not hand-corrected, so names and technical terms may be imperfect. The video is authoritative.

hey how's it going his cute Towser from the CTO visor calm but today CTO daily dose for october 30th 2017 first off head tip to the google app team the Google App Engine team for the concept or inspiration behind this chart they actually tweeted this out via Kelsey Hightower on his Twitter feed on the relationship of functions all the way to bare metal and I thought that walk through this chart is a precursor to tomorrow's CTO daily dose and webinar talking about service in general so

I've struggled with this concept of how is server list different than traditional pass platform-as-a-service ever since the concept of lambda has been introduced if you write cloak cold to a platform and that platform executes code and you don't have to worry about the servers behind the scene then how is passed not serverless so I think this chart helps break down the relationship I think this is a question between both operations and the developer the CNC F had a really great debate between members of the CNC

f working group for service there was folks such as AWS on the line and people from projects Alex Ellis who have won the CTO daily dose shortly who talked through the value of functions as a service versus the concept of service if you take an open source project like open set fast install it in your infrastructure then you don't reap the financial benefits of a lambda if you're not executing code and lambda then you're not charged for compute capacity however if you have an open source

project such as open fast or open whisk or name the open source projects or premises and you don't consume resources you still pay for the servers you also still have the operational burden of the server's themselves so I think that plays nicely into discussing the difference between platform as a service functions as a service as it relates to services so these top two areas the functions the applications the functions you write as part of your application make up apps that's a service view from the developer

perspective so from operations to developers teams perspective the developers are more concerned with this server list layer and which they jet write cold that creates apps they can focus on the apps and the functions that make up the apps those functions and apps can exist in containers which we're now getting into the concept of functionally at the operations later our operations teams are most definitely concerned with containers that can hold functions and this is where the concept of pass start to bleed in these functions can

be written towards a pass platform such as cloud foundry the mechanism to run those functions can be within either a container or virtual machine which is most definitely part of the infrastructure silo so the big question on whether a past is func as a service solution or if it's just a traditional pass with servers behind the scene comes from your operating perspective if this stuff if the infrastructure is operating inside of the cloud then from your organization's perspective your entire organization's perspective is all service your

only interaction with the infrastructure is functions and apps developers write functions and apps none of your operations teams concern themselves with the actual underlying infrastructure so AWS lambda V Marcus officers Kyle functions I think it's called or I think that's that might be Google's platform but their functions as a service solution or even passes so a Cloud Foundry hosted by IBM you wouldn't worry about the underlying infrastructure you just write code to the Cloud Foundry pass and IBM operates the servers behind the scenes however if

you're deploying one of these solutions yourself on premises such as open fast then you are going to concern concern yourself with the docker container level the kubernetes that's handling auto scaling the virtual machine that both kubernetes control plane and docker is running on and of course the bare metal servers themselves that hosts the infrastructure so this is a evolution a start of the conversation at how server list itself the the broader term service is much more about the operating model than it is a specific technology

it's a relative conversation to your environment where you're managing and where what you're consuming I know it's a deep topic we'll talk more about this and tomorrow's webinar on the c2 advisor comm visit me on the web the CTO advisor dot-com to rest there for webinar about server list which I talked about this whole diagram much deeper detail hour-long conversation look forward to having it with you until then talkie on the next CTO daily dose