Reflections
On
Serverless
Diego Pacheco
@diego_pacheco
❏ Cat's Father
❏ Principal Software Architect
❏ Agile Coach
❏ SOA/Microservices Expert
❏ DevOps Practitioner
❏ Speaker
❏ Author
diegopacheco
http://diego-pacheco.blogspot.com.br/
About me...
https://diegopacheco.github.io/
Time goes by...
❏ SOA
❏ APIS
❏ Microservices
❏ Serverless
SOA
SOA
+ Services
+ Abstraction
+ Interoperability
+ Shared Services
+ Evolutionary Ref.
+ Flexibility
- ESB (integration)
- Soap / XML
- Enterprise
- Internal
SOA by Amazon
APIs
API
+ REST + JSON
+ Life Cycles
+ External
+ Biz Models
+ B2B
- Don't FIT well
- for internal Enterprise
- Often Text(gRPC)
- Issues Beyond RPC
- Batching
- Streaming
- Aggregation(GraphQL)
Microservices
Microservices
+ Isolation
+ Anti-Fragility
+ UX-Degradation
+ Scalability
- Infra Cost
- Complexity
- Expensive
- Expensive for
Small companies
Anti-Docker Moviment
- Do We really need docker?
- Containers == Linux
Features
- Recipes Fallacies
- Image Nightmare
- Kernel Panics
- Lack of Cluster(K8s)
- Packing (Mesos , Nomad,
FAT jar)
Cloud Providers OSS Vampires?
- Vampires Suck!
- OSS Makes money by
- SaaS and Services
- Cloud Providers are doing
both and also better :-)
- Its OSS Dead?
- License change? Redis.
Could we Abstract “value”?
- Value from service(utilization)
- Cloud-Native VS Abstractions
- Product VS Project Shift
- IT trying to get control back
- Big Vs Small Companies
- Complexities of Real
multi-cloud
Serverless - Wave 1 - Functions
Serverless - wave 1
+ Abstraction
+ Pay on Demand
+ Event Driven
- Tooling
- Limited AutoScaling
- Not For All kinds of
problems
Serverless - Wave 2 - Multi-Cloud
So many OSS Frameworks on top of k8s
Serverless Framework
http://serverless.com
Open FaaS
https://www.openfaas.com/
Iron Functions
https://github.com/iron-io/functions
Open Whisck
https://openwhisk.apache.org/
Spring Cloud Functions
https://cloud.spring.io/spring-cloud-functio
n/
fn project
https://github.com/fnproject/fn
Google Weight in
Google Vision on Serverless
Google Vision on Serverless
Serverless - wave 2
+ Better Tolling
+ Fast for Small
+ Multi-Cloud
+ No Lock-In
+ PaaS Killer
- Data and Blast
Radius
- Data is not there yet.
- Are we doing a ESB
on wave 3?
Conclusions
❏ Kubernetes is the new Linux but...
❏ Another Layer of Indirection
❏ We are starting from “backend” there is data and more…
❏ We might see an new Cloud ESB soon
❏ Docker value is fading == “packing”
❏ Product vs Project Shift | Small vs Big Companies
❏ OSS licenses might change: facebook / redis
❏ No-*: NoOps, NoBackend, NoServers (reduce layers)
Reflections
On
Serverless
Diego Pacheco

Reflections On Serverless

  • 1.
  • 2.
    @diego_pacheco ❏ Cat's Father ❏Principal Software Architect ❏ Agile Coach ❏ SOA/Microservices Expert ❏ DevOps Practitioner ❏ Speaker ❏ Author diegopacheco http://diego-pacheco.blogspot.com.br/ About me... https://diegopacheco.github.io/
  • 3.
    Time goes by... ❏SOA ❏ APIS ❏ Microservices ❏ Serverless
  • 4.
  • 5.
    SOA + Services + Abstraction +Interoperability + Shared Services + Evolutionary Ref. + Flexibility - ESB (integration) - Soap / XML - Enterprise - Internal
  • 6.
  • 7.
  • 8.
    API + REST +JSON + Life Cycles + External + Biz Models + B2B - Don't FIT well - for internal Enterprise - Often Text(gRPC) - Issues Beyond RPC - Batching - Streaming - Aggregation(GraphQL)
  • 9.
  • 10.
    Microservices + Isolation + Anti-Fragility +UX-Degradation + Scalability - Infra Cost - Complexity - Expensive - Expensive for Small companies
  • 11.
    Anti-Docker Moviment - DoWe really need docker? - Containers == Linux Features - Recipes Fallacies - Image Nightmare - Kernel Panics - Lack of Cluster(K8s) - Packing (Mesos , Nomad, FAT jar)
  • 12.
    Cloud Providers OSSVampires? - Vampires Suck! - OSS Makes money by - SaaS and Services - Cloud Providers are doing both and also better :-) - Its OSS Dead? - License change? Redis.
  • 13.
    Could we Abstract“value”? - Value from service(utilization) - Cloud-Native VS Abstractions - Product VS Project Shift - IT trying to get control back - Big Vs Small Companies - Complexities of Real multi-cloud
  • 14.
    Serverless - Wave1 - Functions
  • 15.
    Serverless - wave1 + Abstraction + Pay on Demand + Event Driven - Tooling - Limited AutoScaling - Not For All kinds of problems
  • 16.
    Serverless - Wave2 - Multi-Cloud
  • 17.
    So many OSSFrameworks on top of k8s Serverless Framework http://serverless.com Open FaaS https://www.openfaas.com/ Iron Functions https://github.com/iron-io/functions Open Whisck https://openwhisk.apache.org/ Spring Cloud Functions https://cloud.spring.io/spring-cloud-functio n/ fn project https://github.com/fnproject/fn
  • 18.
  • 19.
    Google Vision onServerless
  • 20.
    Google Vision onServerless
  • 21.
    Serverless - wave2 + Better Tolling + Fast for Small + Multi-Cloud + No Lock-In + PaaS Killer - Data and Blast Radius - Data is not there yet. - Are we doing a ESB on wave 3?
  • 22.
    Conclusions ❏ Kubernetes isthe new Linux but... ❏ Another Layer of Indirection ❏ We are starting from “backend” there is data and more… ❏ We might see an new Cloud ESB soon ❏ Docker value is fading == “packing” ❏ Product vs Project Shift | Small vs Big Companies ❏ OSS licenses might change: facebook / redis ❏ No-*: NoOps, NoBackend, NoServers (reduce layers)
  • 23.