Please make a detailed video on your AWS architecture on how each and every service is used by Codedamm, Also we would like know the usage patterns and avg costs incurred every month in details. It would really help us in knowing how much it costs to maintain a mid sized startup. thank you
I'm definitely keeping an eye out on your AWS architecture, I have nextjs as my prospect fw but also considering Vercel for my deployment, but with the expected bandwidth we're going to eat, I'd probably be better off to AWS or Azure, tho it's still too much info overload looking at their dashboards
AWS Amplify would be worth looking into, it's a serverless backend framework which you can use many of the AWS services. There are many videos here on yt.
Interesting video. Only thing to clarify would be the terminology about data volume and bandwidth. That's not the same thing but keeps getting mixed up in the video. Bandwidth is the capacity of a network link to transfer data over time, sloppily called connection speed and usually measured in (kilo/mega/giga)bits per second. Data transfer volume on the other hand is the total amount of data transferred over a network link over a certain amount of time and that's typically measured in (mega/giga/terra)bytes per second and the topic at hand here. -edit: and yes, i did watch till the end :P
Nice explanation on why you moved away from Vercel... But you also mentioned that AWS is one of the heavy billers as well... So, what made you to choose AWS over cheaper options (like cloudflare etc)?
Cloudflare doesn't support full Node.js runtime - I'm not sure how easy it would be to incorporate Next.js serverless handlers as CF functions. Plus, we use serverless-next.js template and it doesn't support anything except AWS right now.
@@codedamn The plugin also seems to have some limitations tho, would you know how cumbersome it'd be to just dockerize next js and throw it into a container as a service like Google Cloud Run ?
@Heroe Containerization is easy and would have 100% feature parity with vercel, but the problem with containerization would be that you'd lose (or at least would be hard to implement) first class revalidate parameter support the moment you put a cloudfront CDN distribution in front of it for faster access.
How well are the new nextjs 12 features supported on AWS? Specifically middlewares and edge functions, can these be succesfully ported over to AWS, or will vercel become a bottleneck, to use the new features?
Even using a framework like Amplify which heavily marketed as “less code” is and has been a pain in the a. I thought I won’t have to worry so much about infra, and yet I have to otherwise it won’t work as smooth as advertised
@@wsh4and every aws product like so. Serverless ( for more complex tasks) ebs for maintaining. atleast at the initial stages who is just starting with these technologies (which I'm / i was)
I watched this video to the end! This is interesting. I had no idea. I also wonder how AWS cost would compare to a similar setup on google cloud. As much as I would like to use smaller companies it's hard to beat the giants.
I watched this video till the end! :) This video is very interesting and helpful for me. Please explain more about vercel. If my startup still has small visitor on the website is it possible to stick on the free plan? And what is your backend? Is it fullstack using only Next.js api or what? Thank you. I'm still new in Next.js.
Hey mehul , I could notice a bit of a lag when I clicked on full stack learning path from my dashboard. Is it because of cold starts in lambda functions ? Also eventually will there be an option to use cloudflare workers for getstaticprops ?
Yes, it is because of lambda@edge taking some time to execute this. This should be fixed soon - changing some things on the infra side in the coming days. Good thing is this should happen only once per build because after that the page is SSG and stored in S3 and cached by cloudfront (not served via lambda compute)
You can't run Node.js on Cloudflare workers because workers use the V8 engine instead. They basically run Javascript the same way chrome does on the browser.
Good question! 1. Our build times have increased (almost 1.5x) - primarily because we don't cache previous builds manually right now (it was done automatically with vercel). Need to fix that. Vercel builds used to take 8-12 minutes give or take. Current builds take about 14-18 minutes on GitHub actions. 2. The SSR/SSG on lambdas@edge has a worse performance than what Vercel did (even though Vercel uses AWS too - but probably not lambda@edge). We're waiting on migrating the deployment strategy from serverless-trace-target to the new node-file-tree strategy by Vercel in next 12 (nextjs.org/docs/advanced-features/output-file-tracing) that should improve the performance hopefully.
Same boat here, When you start to exceed the limits of the pro plan prices becomes crazy, just for the bandwith and without considering AWS' free plan the difference becomes huge after you exceed the 1TB pro plan's limit, you start wining money from the first 100GB basically. Same from Netlify but it'd few hundreds of additional GB to become interesting. Have you deployed using the serverless plugin ? It seems promising but lacks nice things like some options with Rewrites or newish nextjs 12 options like middlewares. Would be fine if you could explain you process to switch. Thanks.
If you are bootstraped startup and having heavy usage on computing even AWS wont make sense. End of the day you will be on BareMattle on OVH or Hetzner.
Hey Mehul. It would really be helpful if you could detail the prices that you have to pay per month to the services you use for codedamn in any video. And I watched it till the end.
Cloudflare workers have some limitations - especially with regards to runtime. It doesn't run a full node.js runtime so there would be a lot of shimming requirements
:( Our site is fine since its globally distributed over Cloudfront and lambda@edge, but VPC went down (including backend repo and database) that was in us-east-1. A lesson to add regional redundancy in system
I would guess that vercel is actually using AWS for it's infrastructure. I mean a company like that doesn't usually have its own data centers and AWS is the most cost effective for raw resource power
The thing is when you start paying you are not only paying yourself but also for the millions for free accounts that vercel has. Not a business model that I like.
I've worked mostly with AWS. And rest of the services of codedamn are also on AWS. Google cloud and azure might be equally good, but this is what we could use to get up fastest way possible.
Don't forget that by paying vercel you also support the development of nextjs which offers us a way to make money. AWS gives us no JS framework to make money or to learn to code, only a cheaper price. Take this in consideration also.
Vercel has $300M+ in funding and probably hundreds (if not thousands) of enterprise clients paying $XXXX/month. I'm not saying that we shouldn't use Vercel, but our $115/month plan for vercel is insignificant to them
@@codedamn And GNU/Linux basically gives aws the opportunity to exist and gives us everything for free so just make donations to the LF and the FSF. AWS would be more easily replaceable.
So you dislike Vercel because you had billing issues? That is really not fair considering that AWS billing can get outta hand if no alerts are triggered.
Where did I say I dislike Vercel because of our billing issue? Also did you even watch the video where I mention $55/100GB bandwidth which is ~8x expensive than AWS?
I think the reason why aws doesn't bring any drastic changes to their developer experiences is because they can afford that, they know they'll be getting customers either way (They can change the entire experience and not just incremental but an overhaul), at this point amazon is too big and monopolistic to care about the user experience,lol
AWS sucks! Bloated overpriced crappy platform. That $1,000/mo. You’ll pay the developer to maintain aws, you’ll pay to Vercel and you won’t need to maintain anything. 😅
We have different opinions. I know AWS sucks but anything that gives you that much degree of freedom would suck. There are ways to make it suck less - infrastructure as code and idempotent deployments, correct access roles and serverless architecture goes a long way solving "suck" problem.
@@codedamn Honestly, there should not be AWS engineer existing at any places. If it exists, they should be hired by AWS to fix their terrible, unorganized, complex documentation everywhere.
...and that's okay? Too much beginner content already on internet + it's not that I don't post beginner-friendly content here? Something intermediate/advanced once in a while should be fine right?
![ร้อยพ่อพันแม่ - WanMai [Official MV]](http://i.ytimg.com/vi/XY2pNYASEbg/mqdefault.jpg)
Please make a detailed video on your AWS architecture on how each and every service is used by Codedamm, Also we would like know the usage patterns and avg costs incurred every month in details. It would really help us in knowing how much it costs to maintain a mid sized startup. thank you
will do
@@codedamn it would be great if you could show us how to host next js app in S3.
+١
@@codedamn Thanks 😀
I'm definitely keeping an eye out on your AWS architecture, I have nextjs as my prospect fw but also considering Vercel for my deployment, but with the expected bandwidth we're going to eat, I'd probably be better off to AWS or Azure, tho it's still too much info overload looking at their dashboards
profile/pfp sauce?
The simplicity of vercel is a great selling point
true
Please make a video on AWS architecture
And how can I use AWS for my project
This would be great.
AWS Amplify would be worth looking into, it's a serverless backend framework which you can use many of the AWS services. There are many videos here on yt.
I agree
@@lardosian Amplify seems as pricy as vercel, there is a serverless next plugin althought it has limitations
I watched this video till the end and I am eagerly waiting for the for the architecture videos.
Yes! Yes! Please make a video on how to move from Vercel to AWS. It will be very helpful.
Interesting video. Only thing to clarify would be the terminology about data volume and bandwidth. That's not the same thing but keeps getting mixed up in the video. Bandwidth is the capacity of a network link to transfer data over time, sloppily called connection speed and usually measured in (kilo/mega/giga)bits per second. Data transfer volume on the other hand is the total amount of data transferred over a network link over a certain amount of time and that's typically measured in (mega/giga/terra)bytes per second and the topic at hand here.
-edit: and yes, i did watch till the end :P
Would love to learn how you deploy nextjs on s3
Coming soon!
AWS Amplify ?
@@RonaldDas42 nope. Amplify is also like vercel . I guess I had seen a repo called serverless-nextjs which most of them use
@@codedamn looking forward
Nice explanation on why you moved away from Vercel... But you also mentioned that AWS is one of the heavy billers as well... So, what made you to choose AWS over cheaper options (like cloudflare etc)?
Cloudflare doesn't support full Node.js runtime - I'm not sure how easy it would be to incorporate Next.js serverless handlers as CF functions. Plus, we use serverless-next.js template and it doesn't support anything except AWS right now.
@@codedamn The plugin also seems to have some limitations tho, would you know how cumbersome it'd be to just dockerize next js and throw it into a container as a service like Google Cloud Run ?
@Heroe Containerization is easy and would have 100% feature parity with vercel, but the problem with containerization would be that you'd lose (or at least would be hard to implement) first class revalidate parameter support the moment you put a cloudfront CDN distribution in front of it for faster access.
How well are the new nextjs 12 features supported on AWS? Specifically middlewares and edge functions, can these be succesfully ported over to AWS, or will vercel become a bottleneck, to use the new features?
Looking forward to the architectural video btw! Great stuff
Deploying jam stack applications on AWS are pain... please make an video on how you did that
Will do soon!
Even using a framework like Amplify which heavily marketed as “less code” is and has been a pain in the a. I thought I won’t have to worry so much about infra, and yet I have to otherwise it won’t work as smooth as advertised
@@wsh4and every aws product like so. Serverless ( for more complex tasks)
ebs for maintaining. atleast at the initial stages who is just starting with these technologies (which I'm / i was)
Can you please make a video on cost calculation for AWS?
A rough estimate calculate would do too.
Amazing video. Waiting for the architecture videos.
why not use cloudflare workers ?
I watched this video to the end! This is interesting. I had no idea. I also wonder how AWS cost would compare to a similar setup on google cloud. As much as I would like to use smaller companies it's hard to beat the giants.
In future I like to develop defi 3.0..which one is better Vercel Or aws?
I watched this video till the end! :)
This video is very interesting and helpful for me. Please explain more about vercel. If my startup still has small visitor on the website is it possible to stick on the free plan? And what is your backend? Is it fullstack using only Next.js api or what? Thank you. I'm still new in Next.js.
Absolutely. You should try as much as you can to stay within $20/month plan. AWS is a headache and should only be considered when you outgrow vercel
@@codedamn I see. Thank you for your reply 😀
Thank you, this was an eye opener !
Hey mehul , I could notice a bit of a lag when I clicked on full stack learning path from my dashboard. Is it because of cold starts in lambda functions ?
Also eventually will there be an option to use cloudflare workers for getstaticprops ?
Yes, it is because of lambda@edge taking some time to execute this. This should be fixed soon - changing some things on the infra side in the coming days. Good thing is this should happen only once per build because after that the page is SSG and stored in S3 and cached by cloudfront (not served via lambda compute)
You can't run Node.js on Cloudflare workers because workers use the V8 engine instead. They basically run Javascript the same way chrome does on the browser.
@@jaiv yes eventually the serverless-nextjs will be cloud agnostic.
i knew a video was coming when I saw ur tweet yesterday lol, and here it is!
Moving to AWS feels like such a pain! thanks for sharing your journey and breaking it down for us
How much does your deployment workflow degraded from Vercel to AWS?
How much does it take to deploy for each case?
Good question!
1. Our build times have increased (almost 1.5x) - primarily because we don't cache previous builds manually right now (it was done automatically with vercel). Need to fix that. Vercel builds used to take 8-12 minutes give or take. Current builds take about 14-18 minutes on GitHub actions.
2. The SSR/SSG on lambdas@edge has a worse performance than what Vercel did (even though Vercel uses AWS too - but probably not lambda@edge). We're waiting on migrating the deployment strategy from serverless-trace-target to the new node-file-tree strategy by Vercel in next 12 (nextjs.org/docs/advanced-features/output-file-tracing) that should improve the performance hopefully.
Really useful video. Help me make my decision to choose hosting platform!
Why didn't u chose cloudflare pages?
I watched this video till the end, thanks for sharing your experience
Serverless execution time for vercel pro is 1000 gb hours then extra 55$ per 100 gb hours, how much is it on aws? Vercel is very expensive.
Why did you not switch to Cloudflare + AWS?
Please create a tutorial how to migrate nextjs project to aws architecture
Coming soon!
Thanks For information.... What is your review of linode...
Why doesn;t you consider CloudFlare or Azure ?
Nice explanation. Thanks for the info. I just have one question why not Linode ?
And your architectural changes for AWS?
Thanks for the awesome video, would love a breakdown of how you moved it to AWS and the setup :)
Hello, can you tell me from which playlist is this video ?
Same boat here, When you start to exceed the limits of the pro plan prices becomes crazy, just for the bandwith and without considering AWS' free plan the difference becomes huge after you exceed the 1TB pro plan's limit, you start wining money from the first 100GB basically. Same from Netlify but it'd few hundreds of additional GB to become interesting.
Have you deployed using the serverless plugin ? It seems promising but lacks nice things like some options with Rewrites or newish nextjs 12 options like middlewares. Would be fine if you could explain you process to switch. Thanks.
AWS is only expensive if you're running EC2, and running it 24x7. if you use serverless services it's actually very very cheap.
you mentationed cloudflare for what ?
If you are bootstraped startup and having heavy usage on computing even AWS wont make sense. End of the day you will be on BareMattle on OVH or Hetzner.
Hey Mehul. It would really be helpful if you could detail the prices that you have to pay per month to the services you use for codedamn in any video. And I watched it till the end.
Sure, that's more startup/business related so probably might go on my personal channel than here.
Thank you for this inside! I wonder how AWS Lambda Edge would compare to Cloudflare Workers?
Cloudflare workers have some limitations - especially with regards to runtime. It doesn't run a full node.js runtime so there would be a lot of shimming requirements
I was looking for the code on setting up on AWS, i had alot of issues and error after moving there
AWS just went down. hilarious
:( Our site is fine since its globally distributed over Cloudfront and lambda@edge, but VPC went down (including backend repo and database) that was in us-east-1. A lesson to add regional redundancy in system
Why not cloud flair?
Thanks a lot for this , I am wondering whether you encountered any performance changes in regards to this migration?
Hello sir, can you please make a video explaining how to host next js projects on AWS since I tried to do it once but failed terribly.
NextJs 13 not supported on NextJS serverless AWS
what does data transfer mean ?
The worst part of Lamda Edge is you cant access VPC bounded services, like DocumentDB.
true
also, impossible to keep it warm - with less global traffic you're probably going to spend more time in cold starts than in execution
how much you might in case if you used aws instead of vercel for that 115dollars
Any chance of having an AWS course on codedamn?
In the DevOps learning path, next year
I would guess that vercel is actually using AWS for it's infrastructure. I mean a company like that doesn't usually have its own data centers and AWS is the most cost effective for raw resource power
My question is why not google cloud ui good there i think
Please do a video about your aws architecture of nextjs app
what about digitalocean app platform?
DigitalOcean doesn't have the infrastructure to handle all the parts Next.js needs for SSG/ISR, etc.
The thing is when you start paying you are not only paying yourself but also for the millions for free accounts that vercel has. Not a business model that I like.
need more tutorial on aws deployment , lamdas, s3 and ec2
Why not GCP or Azure or anything?
Something like.. why Codedamn choose AWS among all...
I've worked mostly with AWS. And rest of the services of codedamn are also on AWS. Google cloud and azure might be equally good, but this is what we could use to get up fastest way possible.
Do you want to share how much it is costing now on AWS for your architecture now?
We are waiting the AWS architecture 🎈
I purchased vps with 5 TB bandwidth / month 😁
yss we want nextjs deployment in aws..
Would love to see a AWS tutorial 🙏❤️
Don't forget that by paying vercel you also support the development of nextjs which offers us a way to make money. AWS gives us no JS framework to make money or to learn to code, only a cheaper price. Take this in consideration also.
Vercel has $300M+ in funding and probably hundreds (if not thousands) of enterprise clients paying $XXXX/month. I'm not saying that we shouldn't use Vercel, but our $115/month plan for vercel is insignificant to them
AWS *gives* vercel the architecture to *exist* so that we can deploy on vercel and make money. So....
@@codedamn And GNU/Linux basically gives aws the opportunity to exist and gives us everything for free so just make donations to the LF and the FSF. AWS would be more easily replaceable.
How much does AWS paid?
AWS charges separately for every component - bandwidth, storage, compute, etc.
please anyone can explain cloudflare pages billing, i have read that they are free.
also gcp vs aws in terms of cost.
I WATCHED THIS VIDEO TILL THE END!
So you dislike Vercel because you had billing issues? That is really not fair considering that AWS billing can get outta hand if no alerts are triggered.
Where did I say I dislike Vercel because of our billing issue? Also did you even watch the video where I mention $55/100GB bandwidth which is ~8x expensive than AWS?
cool.. I watch it all
Anyone having trouble with nextjs and twin macro ?
Even I am also thinking to move out of vercel
I watched this video till the end
Mehul watching "You" on Netflix??
The graphic designer is ;)
wach it until the end.
I think the reason why aws doesn't bring any drastic changes to their developer experiences is because they can afford that, they know they'll be getting customers either way (They can change the entire experience and not just incremental but an overhaul), at this point amazon is too big and monopolistic to care about the user experience,lol
Damn....i knew it !
Why aws not Azure or gcp?
because I'm most familiar with aws
@@codedamn make sense
AWS sucks! Bloated overpriced crappy platform. That $1,000/mo. You’ll pay the developer to maintain aws, you’ll pay to Vercel and you won’t need to maintain anything. 😅
We have different opinions. I know AWS sucks but anything that gives you that much degree of freedom would suck. There are ways to make it suck less - infrastructure as code and idempotent deployments, correct access roles and serverless architecture goes a long way solving "suck" problem.
Funny thing happens when the company founder can also code/work on aws - we don't need a AWS engineer just yet :)
@@codedamn Honestly, there should not be AWS engineer existing at any places. If it exists, they should be hired by AWS to fix their terrible, unorganized, complex documentation everywhere.
The reason you get less views and comments is because your content is not at all for beginners
...and that's okay? Too much beginner content already on internet + it's not that I don't post beginner-friendly content here? Something intermediate/advanced once in a while should be fine right?
@@codedamnyou are very correct
second
Hey second
Haha lol 😂
have you tried @flightcontrolhq??
yss we want nextjs deployment in aws..