Видео

Great video. I really like the real-life example provided with identity being used to authorize the write to firehose. Would have been great if everything else was also backed with an example. Could you redo this video with more example scenarios

Oh the same mantra of: old monoliths bad, new microservices good but scary to software engineers, and GenAI taking over everyone’s work, making software engineers even more frightened, and the CFOs still not getting the Cloud pricing models, with Black Friday being just the spike and the rest of the time hardware sitting idle. You know, that playbook is outdated. There’s so much nuance that came out of the interactions with the Cloud infrastructure and pricing models that this video isn’t even trying to scratch the surface of. We’ve been running on AWS for over a decade, but not for the reasons stated in this interview.

Another great thumbnail 😁 Oh and the episode is good, too. img.ruclips.net/user/viRSV0VDBPRGU/maxresdefault.jpg

If you have kubernetes already running then you can also look at the kubernetes based GitHub runners along with KEDA for autoscaling. Beware of DIND limitations.

I just found you guys, I content was very useful. Please keep it up. You guys gained a new sub.

Another great episode of useful material, and another great RUclips thumbnail. Someone is having a lot of fun with those.

I have been using Python with Cloudformation for few years and I have been pretty happy with the experience. Easier to write code, unit tests and ship libraries using SAM cli.

Awesome content! Thanks!

This combo always appeared natural, fast and cheap but there was not many sources on the topic.

What about Golang

was this before the AWS CDK?

How about free tiers, all new account have it’s own “free tier”?

Amazing! I do have a question though. Many people say that cloudwatch is one of the most expensive AWS services. I understand that cloud cost optimization can help... but many just go towards open source tools like OpenSearch (ElasticSearch) or Prometheus... because you don't have pay that much as cloudwatch..??? really don't understand what they are trying to say to be honest.

great episode!

for app config, if you have cold start lambdas you would pay cost per request received for each one one right? which is 0.0008, so if I have 40 lambdas, every time they are loaded will be 40*0.0008, let's say every 15 min my lambdas are loaded, so in 24 hrs will be 3 dolars, and one month, 90 dolars. I think it make senses to use app config if you have a very active application for all the lambdas

Thanks for covering this topic. It's good to learn about new options, otherwise I just keep using what I already know. I'm looking forward to seeing how CFHT evolves.

I'm a big fan of DynamoDB's scalability, especially since I've been at startups where SQL databases failed when our prototype took off and we suddenly had a surge of users. It nearly killed one startup. That said, if the access pattern becomes too complex, you need to combine DynamoDB with something else, like an SQL database. I attempted a "single table design" once. It was challenging enough that I locked myself in a conference room with whiteboards to map it all out. By the end of the day, I had it figured out and started writing it up. The next morning, my boss added a last-minute feature that required a new access pattern, causing the entire single table design to collapse like a house of cards. I haven't attempted such a design since. Now, when I use DynamoDB, it's for simple use cases with a primary key and sort key (like a timestamp). If it gets more complicated, I'll combine it with an SQL database that might have DynamoDB keys for the large datasets. I'd like to at some point work with a network database like Neo4J or AWS's Neptune, but haven't found a case for it yet. I might invent a personal project just so I can have a reason to try it out.

thank u.

This thumbnail is incredible

I don't know who needs to hear this but I have been using dynamodb as primary db in one project and it is total pain in the rear to design db (access patterns and data modeling) for application with a mobile application and dashboard. No matter how much effort you put in, there will be a point where your existing patterns won't be enough and there no easy way to support new ones.

In this episode, we mentioned that you could use a user email as a primary key in DynamoDB. This can work but from a compliance/security/privacy perspective it might a bit of a problem. So in these cases, if you don't have other unique identifiers for the user, you could use a hash of the user email instead (e.g. md5 or sha).

good video, I'm basically still lost though 😮‍💨

OpenAI could understand the speech of the Italia guy? Alora!!

Thanks for the mention 👍

the thing I appreciate a lot is the fact both of you always share the great resource that serves as a foundation of your speech, thanks for always sharing great content

Good one! :)

I got really good at CloudFormation. I wouldn't say I liked it, but felt it got the job done. Moving now to CDK (prefer Python) in the same language the project is using. Also: A show about DynamoDB single table design would be interesting to me.

Very informative episode. ADF (Amazon Deployment Framework) is another interesting framework for creating and managing accounts. It has drawbacks (like not using CKD v1 and being released slowly) but can be integrated with Control Tower and you can define and create pipelines in the CICD account.

Another great episode, as always. Thanks for sharing your knowledge. You mentioned using email aliases (using "plus" notation) to satisfy the distinct email address requirement for each account. It occurred to me that if the inbox to which all the aliases are associated were compromised, an attacker would have a list of every account in the org including each account's unique email address. Enabling MFA on every account is a no-brainer, and ideally the email inbox should have MFA enabled as well, but it does feel like putting all your eggs in one basket. Maintaining multiple distinct email inboxes is certainly a lot of work, but maybe it's worth it? Or is MFA sufficient protection?

*Promosm* ❣️

Excellent explanation

Great walk through and totally agree with the recommendation. Thank you 👍

Straight to the point and very good explanation. Thx for sharing

Thank you. cleared my confusion on these two concepts.

Hello Team, Could you please let me know how to calculate utterances in live streaming audio from transcribe. If this is possible?

Ty!

Thanks for covering this topic. I knew SCPs were a thing, but never really grasped what problem they solved. As a result, I have not been using them in my organization. Now I understand. If you'll excuse me, I have some work to do ...

Terms are used without introduction and without explanation, for example API gateway, load balancer, Cognito authoriser, groups, principle, .... Rather confusing video.

It has been about 10 months since this episode, and I'm wondering if enough AWS services have been implemented to make this worth a look. Has Application Composer improved enough since this episode that some things mentioned here would change? I took a quick look (Jan 2024) and it seems best for API Gateway + Lambda integrated with the usual AWS Services - DynamoDB / RDS / SNS / SQS / Cognito ... I haven't done a test project but from the console, it looks like it is a quick way to start a CloudFormation template. This brings up an interesting question which is when would someone jump to CDK vs. Step Function compose vs. other options. It seems like there is a lot of overlap between the Infrastructure and code options now.

a

@eoin: not sure if I found the reference to the 2 Enterprise Architecture/ Message driven workflows books you mentioned, in the show notes. Could you provide that please

Also what about debugging AWS API requests/responses using the AWS CLI history? 1) aws configure set cli_history enabled 2) Execute a command like "aws s3 ls" 3) Find the id and show the details with "aws history list" and "aws history show"

What about jmespath tools?

AWS dashboard - Crucial part of the video is missing by itself...

Typical claim check pattern..

Thanks this was a very interesting episode. I ran into the problem is loading pandas into a lambda function a few years ago and found it fit just about the zip file limit. Creating docker images for lambda functions had not come out yet. Knowing that pandas could very well be cached makes it even more compelling to try that route now.