Monday, June 14, 2021

Podcast

 




Meet excelent pople

Great social activity

confirmation you self



Availability of Big Data Clouds - Twin Prime - Satish


Twin Prime the closet analogus describe is CDN

  1. content go faster
  2. api work faster
  3. We made some smart decision
    1. Context is understood
CDN solution
  1. caching
  2. compression
Images
  1. Serial
  2. Parallel
Why availability is important?
  1. Trust is number one availability, performance, security
  2. Availability and uptime are very profound for customers
  3. Examples : Kone - High availability
  4. Examples : AirBnB call center on Salesforce

Availability
  1. core pod design - helped scalability
  2. partition data across the pod
  3. scalability - update a pod and replicate across.

Availability Challenges
  1. hardware failure
  2. SAN failure
  3. Broken cable
  4. Servers going down is No longer the issue
  5. 99.99% available is the North star - i don't care your platform is running, but my service is not working.
    1. Why not 100%?
      1. There is nothing which is 100% available
      2. Availability get harder and harder with every 9
      3. Law of marginal returns curve
  6. What is difference b/w 99.99 vs 99.999 means how many min in hand?
    1. 99.99 - 4min 30sec month
    2. 99.999 - 30 sec  a month - ungodly figure
    3. Difference in cost from 99.99 to 99.999 is exponential costly
  7. Durability and availability
    1. Durability - won't lose the data - 99.99999
    2. availability - you won't lose access to data
  8. Blast radius
    1. Roleout changes in small and propagate to others
    2. contain and scope them 
    3. Safe change
    4. Purple flock - may 29 salesforce down for extended period of time.
      1. Learning?
        1. We know them, but we need to follow them and put in software
        2. Default patterns - easy thing to do is correct thing to do
      2. Right Speed - Not too fast , not too slow

  9. Hardware failure?
    1. SF got good at Hardware not going down when there is hardware failure.
    2. Highly available systems running on top less highly available systems
    3. make redundancy, Software would fail over - hot sand by 
  10. Hbase - bidiretional replication - vagabond project
    1. 99.9 -> 99.95 
    2. Agility 
  11. Falcone - 
    1. Default AWS
    2. BlackJack
    3. Alibaba
    4. GCP

Advantage of Salesforce
  1. very malleable and easy to push changes

Source 
  1. Availability of Big Data - link


Podcast : Shan Appajodu : SDLC Chapter - Security, Agility, Insights, Quality and Intelligence

Left Shift - identify the issue earliers

SDLC - he should not sit staring at screen to check if things are completed
Inner sourcing
SCT - Secured Chain on Trust - everthing is secured
  • Source code system are controlled
  • Code Reviewed
  • How artifacts are build and secured and not having malware
  • Also validate the change
Solar winds - link
Tooling First - makes life easy and reduces the contacts to people

CD - Continuous Delivery / Deployment
ex : Spinnaker - Netflix - all run book in cloud, multi Cloud substrate support 
1. Infrastructure as code
2. Deploy as immutable code - VM, Docker and deploy in Falcon - No patching in Falcon - new deployment.
3. Make declarative deployment.



No comments:

Post a Comment