Serverless is FaaS-tastic - All Things Open Meet-up

TriggerMesh – Multicloud Serverless Management Platform
Serverless is FaaS-tastic

Mark Hinkle, Co-founder TriggerMesh
● Co-founder TriggerMesh, Serverless Management
● Executive Director, Node.js
● VP of Marketing, Linux Foundation
● Head of Citrix Open Source Office
● VP, Cloud.com (acquired by Citrix)
● VP, Zenoss (drove revenue from $400k to $20 mil)
● Director Tech Support, MindSpring (ran a $90 million
BU)
● Governing Boards -, Xen Project, Apache CloudStack
● Editor-in-Chief - Linuxworld, Magazine, Enterprise
Open Source Magazine
Twitter: @mrhinkle
LinkedIn
mrhinkle@triggermesh.com

Busted Old Joint, New Hotness
Monolithic Applications Cloud Native Applications

Example Monolithic Architecture
Data Access
ServiceServiceService
Model View ControllerJavaScriptHTML
Relational
Database

Example Serverless Architecture
Cloud
Native
Application
User authentication
Via Oauth, Identity and
Access Management (IAM)
Static content suchs as HTML,
JavaScript, CSS from web
servers and/or cloud object
storage.
Serverless functions running
compute on serverless clouds
(e.g. Lambda, GCF, ACF) or on-
premise FaaS.
API Gateway
Database(s) - Cloud-based
NoSQL, SQL or on-premise

Serverless is the future (sort of)

Continuous integration, continuous
deployment is going to change and that’s
where things like microservices, containers
are playing a massive role in the outer loop,
but one of the things that I think is going to
completely change how we think about logic
is serverless.
Serverless computation will fundamentally,
not only change the economics of what is
backing computing, but it’s going to be the
core of the future of distributed computing.
So this application paradigm shift of
intelligent cloud and intelligent edge is going
to be pervasively changing everything we do
inside of Windows, inside of Office 365,
inside of Azure and the rest of what you will
hear about in the next three days is about
unpacking this.
Satya Nadella
Keynote - Microsoft Build 2017 conference

The torrid pace of adoption and innovation in
the serverless (Lambda) space has totally blown
us away,
In particular, Lambda, AWS’s main serverless
service, has “grown like crazy,” with hundreds of
thousands of active customers using it in the last
30 days. That’s 300%-plus year-over-year
growth.
“When we launched Lambda, the first serverless
compute service, it was a watershed moment,”
...large generations of customers will skip
instances and containers and go right to
serverless -- in fact, if Amazon.com were
starting today, it would go serverless.
Andy Jassy
Amazon Web Services Inc.
Chief Executive
Re:Invent 2017

Serverless - Believe the hype
Serverless
hype will
plateau
between 2019
and 2022
according to
the Gartner
Hype Cycle

Serverless Market $19.84 Billion by
2025

81% of Enterprises are Multi-cloud

AWS Growth Rate in Enteprise

Automation and Service Trends in AWS
In 2017 Tim Wagner, General Manager of
AWS Lambda mentioned that AWS saw an
impressive trend of Enterprise adoption
among AWS Lambda specifically in the
Automation, Integration and Management
Services. These are key areas that
TriggerMesh provides a solution to.
ServerlessConf 2017
At Re:Invent 2018 Serverless was the
heart of Amazon’s strategy

Top Technologies on Serverless
Roadmaps for Next 18 Months
The serverless cloud
landscape is multicloud.
TriggerMesh is in a unique
position to provide
multicloud capabilities in
the form of deployment,
migration, and cross cloud
event triggers to help
users maximize their
efficiency and give them
options to prevent lock-in
and increase productivity.

Modern architecture is serverless

What is Serverless?
● Abstraction of backend infrastructure completely
● Execution environment for single purpose functions
● Hosted in public cloud or on-premise
● Serverless functions have a runtime in a stateless
container
○ E.g. Node.js, JavaScript, Go, Python, Java
● Event-driven startup triggers/instant scale out or in.
● Micro-billing instead of per-hour/month billing.

What is Function-as-a-Service (FaaS)?
● Functions-as-a-Service infrastructure delivers serverless computing (Mark’s
definition)
● Abstraction of backend infrastructure completely provide the ability to
deliver serverless computing
● Popular Open Source Serverless Delivery Software
○ Firecracker (Amazon)
○ Fission (Platform 9)
○ Fn (Oracle)
○ Knative/Kubernetes (Google)
○ Kubeless (Bitnami)
○ OpenFaaS (VMware)
○ OpenWhisk (IBM)
○ Serverless (Serverless Inc.)

What is Backend-as-a-Service (BaaS)
BaaS or Backend-as-a-Service refers to the infrastructure
components managed by a cloud provider.
Examples of BaaS
○ AWS S3
○ AWS DynamoDB
○ Azure Kubernetes Service (AKS)
○ Google BigQuery

Event Driven Architecture
● Design pattern around the production and reaction to
events
● Functions are triggered by events
● Examples
○ A file uploaded to an S3 bucket
○ Inserts on a DynamoDB table
○ A message published to an SNS/PubSub topic
○ A CloudWatch alert

Event Example
Notify users when something interesting happens

Serverless Providers
● Amazon Lambda
● Microsoft Azure Functions
● IBM Cloud Functions
● Google Cloud Functions
● Alibaba Function Compute

What are microservices?
● Single purposes functions that
deliver accomplish a single task
● Can run on a server, container
or in a serverless infrastructure
● Microservices are typically
combined to deliver a cloud
native application via
presentation layer
Example of Microservices in a Storefront Application
Account
DBMobile
Application
Browser
API
Gateway
Storefront
App
Inventory
Service
Account
DBInventory
Service
Account
DB
Inventory
Service

Simple Serverless Function
Mobile phone takes
picture and uploads it
to object storage in
the cloud.
Various side photos
are uploaded to an S3
bucket on AWS
Resize images to thumbnails in a photo sharing service
S3 Event triggers an
Amazon Lambda
function.
Amazon Lambda
function resizes
image.
Thumbnails are
written to another S3
bucket on AWS.
Thumbnails are
server to a mobile
application from a
webserver and s3.

Cloud Native Application
Mobile phone
takes picture and
uploads it to
object storage in
the cloud.
Various side photos
are uploaded to an S3
bucket on AWS
Photo sharing service
Thumbnails are
written to and served
from object storage on
AWS S3
Thumbnails are
server to a mobile
application from a
webserver and s3.
User plan
supporting quotas
is queryable via
Salesforce
Alerts users in a
network that new
images from their
friends are available.
Photos are shared
among a social
network of users.
Amazon Lambda function
resizes an image every
time a new

Where serverless is a good idea

Criteria for using serverless
● Asynchronous, concurrent, easy to parallelize into independent units of
work
● Infrequent or has sporadic demand, with large, unpredictable variance in
scaling requirements
● Stateless, ephemeral, without a major need for instantaneous cold start
time
● Highly dynamic in terms of changing business requirements that drive a
need for accelerated developer velocity
Source: CNCF Serverless Whitepaper v1.0

Use Cases for Serverless
1. Multimedia processing
2. Database changes or change data capture
3. IoT sensor input messages
4. Stream processing at scale
5. Chat bots
6. Batch jobs scheduled tasks
7. HTTP REST APIs and web apps
8. Mobile back ends
9. Business logic
10.Continuous integration pipeline
Source: CNCF Serverless Whitepaper v1.0

Serverless Benefits
● Reduced cost
● Risk mitigation
● Increased flexibility of scaling
● Shorter lead time

Autodesk - Increases time to Value
● New accounts are now available in 10 minutes instead of two weeks, the process is
about 99 percent less expensive than before.
● AWS Lambda and a serverless architecture keep other costs down, too. ”Building
Tailor on a perpetual infrastructure alone would have cost us as much as $500 more
each month,” says Williams. “The operational costs of applying security patches and
maintaining configuration management of the servers would have cost at least
another $100 a month. By using a serverless architecture based on AWS Lambda, we
barely exceed $100 a month for all the functions that support Tailor across all three
AWS Regions we support.”
● Security is now stronger too because Tailor ensures uniform configurations and
controls, provides visibility into account security profiles, and augments the Autodesk
SecOps Team’s ability to identify development teams whose servers show
vulnerabilities. “With Tailor on top of AWS Lambda, we can look across all our
accounts, easily understand our security profile, and rest assured that what’s in one
account is in all accounts,”
As the number of AWS accounts in use at Autodesk grew, it became increasingly clear that the
account-creation process was a bottleneck. Autodesk architect, Alan Williams began envisioning
a solution that could automatically set up new AWS accounts and update existing ones.
Case Study - https://aws.amazon.com/solutions/case-studies/autodesk-serverless/

Where serverless falls short
Source: The New Stack Serverless Survey 2018
Q: What are the top three areas in which serverless falls short of expectations? n=251
https://thenewstack.io/ebooks/serverless/guide-to-serverless-technologies/
According to the New Stack’s 2018
Serverless Survey these are the
places where serverless falls short.
TriggerMesh can help provide
solutions to all of these problems
especially the top problems of
portability, control, and risk
reduction by providing a full
serverless lifecycle management
solution that is consistent with
DevOps culture and best practices.

Serverless Gotchas
Portability
● Serverless clouds have potential for tight lockin (e.g users cannot move from AWS to Google and
Azure)
● Functions require rewrite to run on other clouds
Automation / DevOps
● Continuous integration and deployment of function needs integration with version control, current
Cloud offerings are not well integrated with GitHub/GitLab and Bitbucket
● Automation pipeline should be triggered via events and execute functions
Cross-Cloud Communication
● Users want choice and use services in different cloud, leaving the need for cross-cloud eventing.
● No good cross-cloud “event bus” for triggering functions (e.g. Salesforce Heroku to Amazon)

The Cold Start Problem
● Applications that haven't been used recently take
longer to startup and to handle the first request.
● Because serverless happens on use, there aren’t
dedicated instances ready to handle requests
● Solution: Run a function in a dedicated container/VM,
not serverless

TRIGGERMESH PROVIDES A PLATFORM TO MANAGE THE WHOLE
SERVERLESS PRODUCT LIFECYCLE VIA A HOSTED PLATFORM.
HTTP://CLOUD.TRIGGERMESH.IO
- Last Updated - 2/5/19

TriggerMesh Architecture

TriggerMesh Product Features
• Software-as-a-Service
cloud.triggermesh.io
• Source Control Integration
Github/Gitlab/Bitbucket
• Deployment
Kubernete/Knative, Amazon Lambda, Microsoft Azure,
Google Cloud Functions
• Cross Cloud Event Bus
Compatible with all clouds. Today there is no cross-
cloud serverless event bus.
• Function Store
Allow users to share function to build a community
hub for collaboration
• Serverless Framework Integration
Utilizes serverless deployment manifest in the repos
containing the function to deploy in a versatile
manner
• Kubernetes Builds for Versioning
Using Google’s Knative platform, use the Build
extension to provide immutable and reproducible
builds of functions images.
• Cost Optimization
Via observability (open tracing in Istio), monitor time
spent in functions and render cost budget per
function
• Monitoring
Time-series monitoring via Prometheus, defacto
standard in cloud monitoring

Web - www.triggermesh.com
Twitter - @twittermesh
Email – info@triggermesh.com
Mark Hinkle
mrhinkle@triggermesh.com
919.522.3520
Sebastien Goasguen
sebastien@triggermesh.com
+41 79 367 3825

Serverless is FaaS-tastic - All Things Open Meet-up

More Related Content

What's hot (17)

Similar to Serverless is FaaS-tastic - All Things Open Meet-up (20)

More from Mark Hinkle (20)

Recently uploaded (20)

Serverless is FaaS-tastic - All Things Open Meet-up