Azure Kubernetes Service (AKS) High Availability
Вставка
- Опубліковано 25 лип 2024
- Deep dive into thinking about resiliency with Azure Kubernetes Service (AKS).
00:00 Introduction
00:40 Key terms (RPO, RTO, RLO, SLA)
03:15 Think end-to-end
04:52 Availability constructs in Azure
11:22 What do we need for container workloads
16:50 Container registry
22:00 Kubernetes cluster
24:12 Node pools
29:40 Multiple cluster reasons
37:07 Node resiliency
48:30 Deployment configurations and replicas
54:00 Thinking about state
58:30 Levels of resiliency
1:04:33 Close - Наука та технологія
3 years down the line, still one of the best videos on HA/DR in Kubernetes.
Great job again John! Thanks for all the effort you put into your videos!
Thank you very much for your consistent, high quality and informative content John! It's really helped me a ton over the years to understand Azure!
I appreciate that! Thank you
merci John. Super high quality, very pro, very instructive.
Excellent AKS stuff once again! Very much appreciated
Blimey, AKS has been such a black box to me and I’ve not been arsed to try and learn it, but looking to do the 303 and I know it’ll come up - as always you’re ability to break it down and take the scary bits away is incredible. Thanks again, John!
Glad was helpful. Good luck!
John, thank you so much MAN , you don't know how much you helped me in the video.
looking forward to watch others. appreciate your way of explanation its perfect 👍 and make things easy to understand.
Very welcome
Excellent video asusual , Its really awesome how you think through every possible scenario 👍🔥 .
I appreciate that! Thank you!
Again, another perfect stuff!!
Thanks!
Great as always, thank you!
Glad you enjoyed it!
Perfect John!, thanks. AKS force.
You bet!
Super Great Staff, Man. Thanks!
Welcome
Great stuff!
Thanks!
John you Rock
this helped a lot!!
Great!
Thank you John... Perfect stuff.. #RESILIENTAKSENV
Hi John!
Thanks for the great work!
Is there any chance I can get some more details about 41:05 issue with managed disks? Because it may be our future use scenario and I want to be prepared for possible issues
Basically you just need to ensure you evenly scale across zones as a managed disk only can be used in the zone it’s created. The risk is you try to scale but can’t in one AZ but could in another and that other couldn’t mount the disk. It’s edge scenario.
Hey John, how do you prepare for a video like this and avoid going on tangents/rambling? Do you have outline/slides, and or do practice run thrus? I'd be curious to hear how you spend your time prepping for these and how long that prep takes.
It really depends. If it’s whiteboard only then normally I do a draft drawing of how I think it makes sense to approach pretending I’m explaining it to someone so I can see what builds the best. Sometimes I’ll have a OneNote to remind me of key points to cover. I do ramble sometimes :)
Hey John, content is always top notch. But can you tell us where do you get these cool t-shirts?
lol, honestly varies. i just look around. sometimes amazon, sometimes other places.
Thanks a lot John for the great video.
I have a query..
If we create a single cluster AZ aware and lets say we have a POD with persistence storage on Node 1 in Az1 and if the same POD gets schedule on Node 2 in Az 2,will it be able to map the same persistence storage in AZ1 , if not then what would be recommended solution.
No it wouldn’t which is what I was talking about potential reasons to create zonal aks :)
Note aks is generally smart enough to recreate pod in same fault domain if pvc if it can
@@NTFAQGuy Thanks John
John, I didn't catch how can I see in which regions I can deploy AKS with AZ enabled? Because as I see all my clusters have no AZ enabled, and as I understand I would benefit from it?
You need to check if the region supports AZs. Lookup the regions on azure.com and it shows if AZs are supported or not
This is awesome stuff, I just have 2 queries 😁.
1. If my AKS pods have Azure files with GRS PVs. If I am trying recreate everything using arm jason , manifest yaml in different region, how my PODs pvc will point to the existing Azure files PVs which has state and not try to generate new PV?
2. For HA active active setup can pods from different regional AKS clusters can be pointed to use same Azure Files GRS PV?
I don't believe the PVC would link to an existing share if you recreated. There would be manual steps required.
Different regions would be different AKS clusters and therefore would not share PV/PVC. GRS is also only available in one region for Files. Again you would require other steps to share to my knowledge.
Hi John, amazing video. Where can i find the whiteboard for this lesson?
Not available.
thank for your video John!!! need your help to start AKS as im in beginner state could you please provide the route to get start AKS.
Thanks,
I have an AKS overview video as well which may help.
@@NTFAQGuy can we have start to end video for aks
Hi John,
Quick question.
Public IPs are not allowed in my company, all our apps using WAF. and most of the services we connect to we use private points.
I'm trying to deploy an AKS with CNI. and by default when creating the AKS cluster it creates a load balancer with public IP address!
Is there a way to choose an internal LB instead of the public that Azure creates for AKS ?
Thank you in advance
i cover that in the AKS networking deep dive. yes :-)
@@NTFAQGuy Awesome, thanks and appreciate you quick response
how to achieve regional resiliency in azure kubernetes service
You deploy an instance per region then balance between using cross region balancer (just did video on this) BUT you do need to consider database between region replication etc. I have videos on dr and resilience where talk about that.