De nos jours, personne ne nie l’importance de l’observabilité surtout lorsque l’on parle de systèmes distribuées en production. Il est de plus en plus difficile de comprendre et d’avoir une vue réaliste de l’état interne de notre système dans un tel contexte, avec des déploiements de plus en plus fréquents et une augmentation des interactions entre les services. D’autant plus que nous devons faire face maintenant à l’arrivée du Serverless avec une granularité à la fonction et une multitude de services 100 % managés.
2. Ippon en quelques mots.
16ans
2002 - 2018
360
Collaborateurs
38M€
CA
2018
48M€
Prévisionnel
2019
4
Continents
Cabinet de conseil Digital · Cloud · Data
Nous croyons en nos équipes pour donner du sens et toute son essence à la technologie.
IPPON 2019
3. IPPON 2019
#Steve HOUËL
JHipster contributor https://www.jhipster.tech/
daSWAG creator https://github.com/daswag
- When not coworking, I am riding my unicorn
in my swimming pool
Mes publications
Les architectures Serverless How to do Serverless Livre Blanc Kit de survie Serverless
Senior Cloud Architect
Global Tech Leader
@SteveHouel
4. IPPON 2019
#TourDeFrance #Serverless.
➔ 7 Mai Article Observability pillars applied to Serverless
➔ 9 Mai - Bordeaux Observability pillars applied to Serverless
➔ 13 Mai WhitePaper Serverless
➔ 14 Mai - Nantes To be or not to be Serverless
➔ 21 Mai - Toulouse Observability pillars applied to Serverless
➔ 23 Mai - Lille To be or not to be Serverless
➔ 29 Mai - Paris Observability pillars applied to Serverless
➔ Juin Magazine Programmez.com - Architectures Serverless de la
théorie à la pratique
➔ 11 Juin - Lyon Observability pillars applied to Serverless
➔ 19 Juin - Lille Serverless et sécurité, ce qu’il faut comprendre
9. IPPON 2019
Abraham Wald
Wald noted that the study only considered the
aircraft that had survived their missions—the
bombers that had been shot down were not present
for the damage assessment.
The holes in the returning aircraft, then, represented
areas where a bomber could take damage and still
return home safely. Wald proposed that the Navy
instead reinforce the areas where the returning
aircraft were unscathed, since those were the areas
that, if hit, would cause the plane to be lost.
source: https://en.wikipedia.org/wiki/Abraham_Wald
10. IPPON 2019
Abraham Wald
Wald noted that the study only considered the
aircraft that had survived their missions—the
bombers that had been shot down were not present
for the damage assessment.
The holes in the returning aircraft, then, represented
areas where a bomber could take damage and still
return home safely. Wald proposed that the Navy
instead reinforce the areas where the returning
aircraft were unscathed, since those were the
areas that, if hit, would cause the plane to be
lost.
source: https://en.wikipedia.org/wiki/Abraham_Wald
14. IPPON 2019
Solving questions?
Pictures are loading slowly for some users. Why?
(LAMP Stack)
App Tier capacity exceeded
Connections to the database
are slower than normal,
causing connections to
timeout, and latency to rise
21. IPPON 2019
Observability definition
In control theory, observability is a measure of how well
internal states of a system can be inferred from knowledge
of its external outputs
Source: https://en.wikipedia.org/wiki/Observability
34. IPPON 2019
What is Serverless?
● No need to manage servers
● No need to even think about servers
● No need to provision infrastructure
● Pay only for what to use
● Use backend as a Service (BaaS)
● Deploy functions, not apps (FaaS)
● It’s event based/oriented
Well… It means no servers right?
35. IPPON 2019
Drawbacks
➔ Loss of Server optimizations
➔ Implementation drawbacks:
● Execution duration
● Startup latency
● Tests
● Packaging / versioning
39. IPPON 2019
Higher concurrency to observability system
Lambda
Function
OBSERVABILITY
SYSTEM
(IaaS)
Function
Function
Function
Function
Function
Function
44. IPPON 2019
Datadog
SaaS product, created in 2010 by 2 French,
started as an Infrastructure Monitoring Tool
● system / application / business metrics
● distributed tracing
● log management
● browser testing and synthetics
● full grown alerting engine on top of that
A LOT of integrations, A LOT of opensource
code
The holes in the returning aircraft, then, represented areas where a bomber could take damage and still return home safely. Wald proposed that the Navy instead reinforce the areas where the returning aircraft were unscathed, since those were the areas that, if hit, would cause the plane to be lost.
The holes in the returning aircraft, then, represented areas where a bomber could take damage and still return home safely. Wald proposed that the Navy instead reinforce the areas where the returning aircraft were unscathed, since those were the areas that, if hit, would cause the plane to be lost.
The holes in the returning aircraft, then, represented areas where a bomber could take damage and still return home safely. Wald proposed that the Navy instead reinforce the areas where the returning aircraft were unscathed, since those were the areas that, if hit, would cause the plane to be lost.
The holes in the returning aircraft, then, represented areas where a bomber could take damage and still return home safely. Wald proposed that the Navy instead reinforce the areas where the returning aircraft were unscathed, since those were the areas that, if hit, would cause the plane to be lost.
The holes in the returning aircraft, then, represented areas where a bomber could take damage and still return home safely. Wald proposed that the Navy instead reinforce the areas where the returning aircraft were unscathed, since those were the areas that, if hit, would cause the plane to be lost.
Change Lamp Stack logo (issue on Pearl) - Perl
Ippon 2018
Observability principles
250+ integrations
over 1,000 employees
over 7,500 customers
Airbnb, Salesforce, Target, AllScript, The NY Times, Hearst, Pivotal
Datadog collects trillions of data points per day
Pas de magie
How to integrate datadog to your AWS env
Deploy Datadog lambda
250+ integrations
over 1,000 employees
over 7,500 customers
Airbnb, Salesforce, Target, AllScript, The NY Times, Hearst, Pivotal
Datadog collects trillions of data points per day
Show Lambda default metrics
How to add custom metrics (extract Billing, memory used)
Deploy custom metrics lambda and plug it to current lambda function
Only plug metrics on datadog iam roles
Show Lambda default metrics
How to add custom metrics (extract Billing, memory used)
Deploy custom metrics lambda and plug it to current lambda function
Only plug metrics on datadog iam roles
Activate Datadog APm on Lambda
Add XRay to our Lambda function
Add Xray action to datadog iam role
See on Datadog
Show how to implement Logs on Lambda
Add Logs to datadog iam roles
Show Logs on DataDog -> PBaillet and what we can do with that