They do… have you ever been certified at all or taken and internal training? Literally the first bullet from the lambda landing page is “Run code without provisioning or managing infrastructure. Simply write and upload code as a .zip file or container image.”
Serverless biggest plus is scale to zero cost, the managed infrastructure is table stakes. Use little pay small, that's what it is about and what ultimately makes it interesting.
I totally agree with your definition of serverless. If it can't scale down to zero, then it's not serverless. One service that does it right is DynamoDB, where they have the on-demand capacity option that does scale to zero, while also having the option to have provisioned capacity where you pay a flat fee for whatever you provisioned.
That sounds necessary but not sufficient. You can scale down ec2 instances to 0 depending on traffic configured by an ASG but I doubt anyone would call it severless.
I’ve just recently found out in the hard way that opensearch serverless is not that serverless. I forgot to destroy a cluster at work and at the end of the month it sent the company back 1 thousand dollars.
When I thinking about serverless, i think that your code doesn't run all the time, it only runs per request. So you can't store things on memory that lives longer than a request, without stuff like redis. Never thought about it based on cost. And edge is distrubuted serverless. If you don't have code running on a server continuously, then it's serverless. Then it's just file hosting and per request functions.
On the contrary of misleading marketing term "serverless", your that kind of videos are priceless, while trying to be a better Dev. Cloud costs are becoming bigger problems in any firm that I can hear from.
Had a bunch of reinvent recaps from AWS today. All the AWS SAs that presented emphasized “…and this is serverless, which means you don’t have to provision the infrastructure…”. So that is their definition.
One of the many examples in our industry where when not being accurate enough at the beginning of defining something, the industry just expand it in all directions, and at the end, everything is both right and wrong, to the point it makes it worse to use the term than not using it.
To be honest, no-one defined serverless before AWS basically come up with Lambda, and then it was coming mostly from AWS. So, it is really up to them how they are going to define it and this definition can surely change. It is already for a long time that Serverless is only about scalability/elasticity and not bothering about infra setup and paying per some units, not servers.
Serverless ain’t what it used to be - that is for sure. It’s very disappointing and I doubt it will change. The good thing is that many of the key components that are required to build apps in the cloud are the original (and truly) serverless ones.
I'm really serverless, I integrated the VM server into my $349 desktop and saved a lot of money. Hardware: Ryzen 3 2200G; 16GB; 512GB nvme; 2TB HDD with a 128GB SSD cache. Software: Host a minimal install of Ubuntu 23.10 on OpenZFS 2.2.0 and Virtualbox 7.0.12. VBox VMs: Xubuntu 22.04 LTS (Communication; WhatsApp; Email; etc); Ubuntu 16.04 ESM (Banking) encrypted by VBox with the newest snaps for Firefox and LibreOffice(Calc); Ubuntu Unity 23.10 (MultiMedia); Ubuntu 23.10 (Experiments); Windows XP Home (Jukebox) and Windows 11 Pro (Just in case).
Thanks for covering this... It never made any sense to me because you literally need to interact with a server to receive the results of the server less functions/services...
Excellent video!... I agree with you, however is important to see that the services mentioned to present a model stateful such as are the databases engines in the other hand AWS overhead the term "serverless" for so many things; its a bad branding strategies
The new "serverless" announcement I was excited for was elasticache. Seems like something that would be easy to scale to zero, nope. If I understood the pricing it is a minimum of .5 GB so $45 a month plus compute.
I'm a big fan of your work and the way you explain things in an easy manner. I wish you could have a course on developing real serverless solutions using CDK
I totally agree "serverless" it's not always the best solution and when you use it it's important check cost, in my company we have hundreds Lambda with thousands running every day that means hundreds euro every month... obviously AWS try to sell serverless like a solution of all problems but it's not the right!
Serverless isn’t about replacing servers/containers, it’s about not having to hire the people that manage them. The price you pay for a 24/7 Kubernetes team buys a LOT of Lambda compute.
people complained about serverless taking too long to be up and running and so aws introduced this kind of fake serverless that is not serverless anymore, but just a fancy word for autoscaling...
Well done 🎉 Should we rather call serverless ‘stateless, on demand compute’? And if that is really the case, does it mean it can evolve to run on any registered available device?
They just mean that you don't pick or have to manage the infrastructure. You do pay for capacity and if you want capacity available when not in use then you have to pay.
The standard definition of serverless should be zero compute, zero dollars. If it doesn't fit that definition, in my opinion it's not serverless. I'd love to see the developer community adopt some type of phrase when describing these services like "double-zero serverless" and "non-zero serverless" to help create more transparency.
If AWS can't provide a true serverless option, then I would just use ec2 instance, install what I need for the job and control the up time with a scheduler. Yes, it would require upfront work to set it up, but at least I would be in control of the service.
💯 agree. We run multiple applications for multiple clients on AWS and we automated startup and shutdown of services where it is possible and this saves much more than any serverless solution. Of course AWS would like you to pay more, not less.
I think those real serverless services are mostly a way of inserting AWS services in small small business applications. Serverless has a way of running code, a database, auth, and WebSockets. This is all most small apps needs, so doesn't make sense for AWS to make real serverless anymore, they will just use the serverless branding to make some product sound easier to manager or cheaper.
You hit it on the head. Aurora serverless v2 does not scale down to 0 and i got a bill i wasnt expecting so i ended up moving to a standard postgres rds db inatance and its much cheaper. I know its my fault for not reading the fine print and calculating costs but amazon seems to put it front and center (understandably, they want you to ise it) and it was kind of misleading to me
To effectively sell tech you must have terms that make the consumer feel like he is getting high value for his money. One company that does it well is the one that you'll find using terms like center stage, retina display to describe things that are ordinary and normal but consumers get the idea that they are better and of higher value.
I am an Enterprise Technical Architect, and while I agree with your point of view, I myself have not qualified ‘serverless’ by the ability to scale down to zero usage. I’ve, perhaps incorrectly, have always thought of it as a non-K8, non-IaaC, non-persistent presence within a service provider’s infrastructure. However, from a Admin point of view, I can see how your qualification fits as well. You probably are more technical knowledgeable of AWS than I am, however; I prefer both GCP and Azure to AWS.
It's quite ironic that many company realize that Cloud & serverless is not cheaper and better for them then, having hardware on prem. A client spend 3-4x more in "Cloud Services" then they would if they bought licenses and hardware for on prem in 5y.
My god this comment section is confused. “Serverless” means that there are no servers that you need to personally manage for your software architecture. It has never meant “without a server”. The people getting confused here have clearly never had to manually manage clusters of servers. What a pain in the ass that is. It is worth the extra cost as long as you know how to design efficient architectures.
I think we’re getting too caught up in semantics. To me serverless is about not managing servers (no need to think about subnets, compute capacity, storage capacity, networking, redundancy, and easier DR) and secondly about saving costs. The drawback to serverless is the speed at which AWS can give you capacity fast enough to make the service viable. So AWS has been diluting the concept of scale down to nothing and forcing minimums, like in aurora serverless and passing on the costs to consumers. It’s been the same with provisioned concurrency in lambdas. With this, you get the benefit of the serverless technology (e.g. no servers to manage) with reasonable response times when scaling up from nothing. The bigger problem I have when it comes to semantics is the term “fully managed” when applied to RDS. If RDS is fully managed, what do we call services like DynamoDB or S3 that are truly fully managed: “really, truly, fully managed”?
You: How much do you charge for your service? Amazon: 5 dollars per crondos You: What's a crondo? Amazon: It's a hundred dandros. That's literally what the pricing model is. If this is not a scam, I don't know what is.
To me severless just means not having to manage a server, you only concern yourself with the application itself. Scalping to zero isn't severless, but a consumption model. The consumption model is really only possible within a serverless environment, but is not synonymous. I think the issue here is likely less with what serverless means, as that seems self evident by the name, and more with a associating it with something that it didn't mean. That doesn't mean that association wasn't reasonable at the time, but just there was a misunderstanding and nothing has really changed.
I think serverless just means you don't need to or even get to worry about the server. Like applying system patches. Or worrying about mount points. Our whatever server things you don't think about anymore. That doesn't mean it doesn't actually run on a server, or that it scales to zero
I get your definition and the hopeful thinking, but what they mean by serverless is "(mostly) aws managed resources allocation". The price is driven by the service where AWS takes away the complexity, not how fancy the compute/storage is. It's their world we work in and that's their created term. It has nothing to do with what the customer wants, it's all about what the tech giant knows the customer will pay for.
Hi Daniel 👋 new sub here! My website users upload 3 to 6 json files. I'm considering to save these files in a s3 bucket. Should I save the links to json files in a database like mysql? Your tutorials are so easy to follow, please keep the great content rolling 🎉
If the json files follow a predefined schema, the json files could themselves be in the SQL db. Perhaps a different table linked via a key. You will retrieve info faster, and write less code.
I appreciate your sentiment and completely agree that scaling to zero ought to be an essential component of the definition of serverless. This couples particularly poorly with AWS' frustratingly non self-describing branding of their services, especially compared to its Azure and GCP counterparts. I think it would be prudent to iron out their branding overall, making sure serverless means the same across the board and maybe doing something overall with their service/resource naming convention to append something or alter something to make their naming make sense.
I never really understood the buzzword "Serverless" I mean there really isn't a such thing as being severless when using the cloud. I mean what's a Cloud platform without the physical underlying infrastructure that runs it? Just because you may not manage it, doesn't mean you aren't using physical hardware that runs thr Middleware and everything regardless if it's PaaS or IaaS. You are still using some else's hardware infrastructure.
Serverless just means you do not need to concern yourself with the hardware or OS. You only concern yourself with the application you are running, and let AWS, Azure, etc. handle maintaining the hardware.
@@evancombs5159 That's more like application deployment but again, all of that is STILL running on a physical server infrastructure in a data center regardless if you don't manage it. No different than accessing a website on the web. You are still interacting with a server. The client sends a request to the server and server responds and exchanges the information back.
This is a total piss-take. I agree, Serverless, these things are NOT, and the naming needs changed to relect that. Really angry 😡that this is actually a thing, and it needs nipped in the bud -- it has somehow become "normalised".
Even the original concept and name "serverless" was a bit misleading. It sounds like magic. How does the code run without a server?? Server-on-demand might have been a better name.
At reinvent they had a talk about server-less best practices and in the first 5 minutes touch on what sever-less was 10 years ago and what it is now… ua-cam.com/video/sdCA0Y7QDrM/v-deo.htmlsi=vZHh0d-srZyGR2pV
What ARE you talking about? Serverless is NOT scale-to-zero! People got that idea stuck in their head when lambda came out and worked that way! Serverless means exactly one thing, and _ONE_ thing _ONLY:_ YOU don't rent a server, neither directly, nor as a consequence of running an instance of a service, an RDS database for example. And about the cost: You have to keep an eye on your cost and calculate them ahead of time when running services in the cloud? Shocking development, Sherlock! Seriously, the REAL problem is that with AWS' low-bar-services like Amplify, you get a lot of devs on the cloud who are in WAY over their heads. You need a license to drive a car, but you don't need a license to mortgage your income for the next twenty years in the cloud, because you couldn't be bothered to do the math ahead of time. Maybe what AWS should really implement is a kiddie mode that just pulls the plug at a certain monthly cost and to hell with anything but backups.
Azure has that feature, you set a cost limit and things turn off once that budget is met. It is also my understanding that Azure is generally cheaper than AWS so a lot of this may be less of an issue over there if my understanding is correct.
Serverless or not, when you buy into all this cloud nonsense, don't be surprised when you discover that you are being ripped off by and completely dependent on a big company. Why would anyone "run their reports on Friday" on AWS if this can be done in-house by building your own server? Self-host everything that you need and you will never have to spend time and energy parsing through yet another portion of big tech marketing garbage. And if you can't self-host something, you should not do it at all.
The big reasons are uptime and scalability. Two of the hardest things to do with on-prem is make sure there it's not downtime due to network or server issues, and to quickly scale a server as needed (i.e. have normally unused server capacity). There are other things like location redundancy. These are things even larger companies struggle with, let alone small or medium sized companies.
There is so much more wrong with AWS. It is very hard to find actual specifications, just propaganda everywhere. It is waste of time. Just hire proper sys admin and have your dedicated server running at fraction of cost.
I’m a sys dev at Amazon. They should really just say that you don’t need to manage the infrastructure. Great video btw!
I don't work at AWS but that was my understanding too
@@devonbrazier4855 Numerous courses on skillbuilder are all labeled serverless too lol They are really pushing the terminology
They do… have you ever been certified at all or taken and internal training? Literally the first bullet from the lambda landing page is “Run code without provisioning or managing infrastructure. Simply write and upload code as a .zip file or container image.”
@@Coral_dudeyou don't know what infrastructure is, do you?
Serverless biggest plus is scale to zero cost, the managed infrastructure is table stakes. Use little pay small, that's what it is about and what ultimately makes it interesting.
I totally agree with your definition of serverless. If it can't scale down to zero, then it's not serverless. One service that does it right is DynamoDB, where they have the on-demand capacity option that does scale to zero, while also having the option to have provisioned capacity where you pay a flat fee for whatever you provisioned.
That sounds necessary but not sufficient. You can scale down ec2 instances to 0 depending on traffic configured by an ASG but I doubt anyone would call it severless.
@@josephs.7960 I mean, it's difficult to be serverless when you have to manage servers.
@@josephs.7960well that’s still managing servers, just that it can scale to 0
Lambda does it right too
Thanks for your time bro. I have learnt a LOOOOT with your YT channel. From zero to hero. You're the best
I’ve just recently found out in the hard way that opensearch serverless is not that serverless. I forgot to destroy a cluster at work and at the end of the month it sent the company back 1 thousand dollars.
100 percent right! Actually they could have us pay less than 10$ per quarter when we dont use this serverless products.Great video!
When I thinking about serverless, i think that your code doesn't run all the time, it only runs per request.
So you can't store things on memory that lives longer than a request, without stuff like redis.
Never thought about it based on cost.
And edge is distrubuted serverless.
If you don't have code running on a server continuously, then it's serverless. Then it's just file hosting and per request functions.
On the contrary of misleading marketing term "serverless", your that kind of videos are priceless, while trying to be a better Dev. Cloud costs are becoming bigger problems in any firm that I can hear from.
100% agree with AOSS. It costs like hell and blew my mind to see the bill after 1 week
"Serverless" is just a marketing term at this point
Thanks for the heads up. I really thought serverless meant serverless. I’ll be reading the fine print more closely going forward.
Had a bunch of reinvent recaps from AWS today. All the AWS SAs that presented emphasized “…and this is serverless, which means you don’t have to provision the infrastructure…”. So that is their definition.
Agree- Managed Capacity is a better name. However capacity is a big hot button issue and thing AWS doesn't like to talk about.
I think the most accurate name is "vendor lock-in".
One of the many examples in our industry where when not being accurate enough at the beginning of defining something, the industry just expand it in all directions, and at the end, everything is both right and wrong, to the point it makes it worse to use the term than not using it.
To be honest, no-one defined serverless before AWS basically come up with Lambda, and then it was coming mostly from AWS. So, it is really up to them how they are going to define it and this definition can surely change. It is already for a long time that Serverless is only about scalability/elasticity and not bothering about infra setup and paying per some units, not servers.
I'm surprised they don't even allow an opt in for Aurora Total Serverless (like v1) at the cost of longer cold starts.
Serverless ain’t what it used to be - that is for sure.
It’s very disappointing and I doubt it will change. The good thing is that many of the key components that are required to build apps in the cloud are the original (and truly) serverless ones.
What are you on about? It is easier than ever to deploy serverless architectures.
I'm really serverless, I integrated the VM server into my $349 desktop and saved a lot of money. Hardware: Ryzen 3 2200G; 16GB; 512GB nvme; 2TB HDD with a 128GB SSD cache. Software: Host a minimal install of Ubuntu 23.10 on OpenZFS 2.2.0 and Virtualbox 7.0.12.
VBox VMs: Xubuntu 22.04 LTS (Communication; WhatsApp; Email; etc); Ubuntu 16.04 ESM (Banking) encrypted by VBox with the newest snaps for Firefox and LibreOffice(Calc); Ubuntu Unity 23.10 (MultiMedia); Ubuntu 23.10 (Experiments); Windows XP Home (Jukebox) and Windows 11 Pro (Just in case).
Thanks for covering this... It never made any sense to me because you literally need to interact with a server to receive the results of the server less functions/services...
Excellent video!... I agree with you, however is important to see that the services mentioned to present a model stateful such as are the databases engines in the other hand AWS overhead the term "serverless" for so many things; its a bad branding strategies
The new "serverless" announcement I was excited for was elasticache. Seems like something that would be easy to scale to zero, nope. If I understood the pricing it is a minimum of .5 GB so $45 a month plus compute.
I'm a big fan of your work and the way you explain things in an easy manner. I wish you could have a course on developing real serverless solutions using CDK
I totally agree "serverless" it's not always the best solution and when you use it it's important check cost, in my company we have hundreds Lambda with thousands running every day that means hundreds euro every month... obviously AWS try to sell serverless like a solution of all problems but it's not the right!
Serverless isn’t about replacing servers/containers, it’s about not having to hire the people that manage them.
The price you pay for a 24/7 Kubernetes team buys a LOT of Lambda compute.
people complained about serverless taking too long to be up and running and so aws introduced this kind of fake serverless that is not serverless anymore, but just a fancy word for autoscaling...
Well done 🎉
Should we rather call serverless ‘stateless, on demand compute’?
And if that is really the case, does it mean it can evolve to run on any registered available device?
Great video. The term "Serverless" has been misleading from the beginning.
They just mean that you don't pick or have to manage the infrastructure. You do pay for capacity and if you want capacity available when not in use then you have to pay.
I took serverless to mean that the customer doesn’t have to manage a server.
Makes told sense. Next up is codeless scripts, and 0 byte storage.
The standard definition of serverless should be zero compute, zero dollars. If it doesn't fit that definition, in my opinion it's not serverless. I'd love to see the developer community adopt some type of phrase when describing these services like "double-zero serverless" and "non-zero serverless" to help create more transparency.
There is a word for it, consumption.
Great video! Perhaps you could do a comparison with Azure also? Always good info in your videos, keep up the great content.
If AWS can't provide a true serverless option, then I would just use ec2 instance, install what I need for the job and control the up time with a scheduler. Yes, it would require upfront work to set it up, but at least I would be in control of the service.
💯 agree. We run multiple applications for multiple clients on AWS and we automated startup and shutdown of services where it is possible and this saves much more than any serverless solution. Of course AWS would like you to pay more, not less.
I think those real serverless services are mostly a way of inserting AWS services in small small business applications. Serverless has a way of running code, a database, auth, and WebSockets. This is all most small apps needs, so doesn't make sense for AWS to make real serverless anymore, they will just use the serverless branding to make some product sound easier to manager or cheaper.
Except they are not cheaper. If you need only low capacity then you are better off with a small regular setup.
It's disingenuous marketing, nothing more, nothing less.
Thanks for the vid.
You hit it on the head. Aurora serverless v2 does not scale down to 0 and i got a bill i wasnt expecting so i ended up moving to a standard postgres rds db inatance and its much cheaper. I know its my fault for not reading the fine print and calculating costs but amazon seems to put it front and center (understandably, they want you to ise it) and it was kind of misleading to me
To effectively sell tech you must have terms that make the consumer feel like he is getting high value for his money. One company that does it well is the one that you'll find using terms like center stage, retina display to describe things that are ordinary and normal but consumers get the idea that they are better and of higher value.
I am an Enterprise Technical Architect, and while I agree with your point of view, I myself have not qualified ‘serverless’ by the ability to scale down to zero usage. I’ve, perhaps incorrectly, have always thought of it as a non-K8, non-IaaC, non-persistent presence within a service provider’s infrastructure. However, from a Admin point of view, I can see how your qualification fits as well. You probably are more technical knowledgeable of AWS than I am, however; I prefer both GCP and Azure to AWS.
great vid. serverless emr, etc. migration back to on prem in our future
Can you do a video that covers the true serverless services or options to scale down to 0
"Managed Servers" is most suitable name.
It's just a managed service which costs you as much as it would to have someone internally manage such a service.
A wise man once said, 'It's not cloud, just someone else's computer'.
It's quite ironic that many company realize that Cloud & serverless is not cheaper and better for them then, having hardware on prem. A client spend 3-4x more in "Cloud Services" then they would if they bought licenses and hardware for on prem in 5y.
Good video. I had this experience with aws's 'serverless' database, and was kinda mad b/c to the deception
+1 to managed capacity
My god this comment section is confused. “Serverless” means that there are no servers that you need to personally manage for your software architecture. It has never meant “without a server”. The people getting confused here have clearly never had to manually manage clusters of servers. What a pain in the ass that is. It is worth the extra cost as long as you know how to design efficient architectures.
GCP is way more cost effective. 3:04 3:06
Aurora serverless v1 can still be spinned up via cloudformation and cdk atleast
I think we’re getting too caught up in semantics. To me serverless is about not managing servers (no need to think about subnets, compute capacity, storage capacity, networking, redundancy, and easier DR) and secondly about saving costs.
The drawback to serverless is the speed at which AWS can give you capacity fast enough to make the service viable. So AWS has been diluting the concept of scale down to nothing and forcing minimums, like in aurora serverless and passing on the costs to consumers. It’s been the same with provisioned concurrency in lambdas. With this, you get the benefit of the serverless technology (e.g. no servers to manage) with reasonable response times when scaling up from nothing.
The bigger problem I have when it comes to semantics is the term “fully managed” when applied to RDS. If RDS is fully managed, what do we call services like DynamoDB or S3 that are truly fully managed: “really, truly, fully managed”?
Serverless sounds like a speedrunning category
The biggest problem is that it costs a lot more than just having your own servers unless you’re really clever about it.
Serverless = Server On Demand?
You: How much do you charge for your service?
Amazon: 5 dollars per crondos
You: What's a crondo?
Amazon: It's a hundred dandros.
That's literally what the pricing model is. If this is not a scam, I don't know what is.
Select Fields from Table Where FilterField like ‘%bullshat%’….YOU owe me 700 bucks buddy! I’ll be expecting that each month.
To me severless just means not having to manage a server, you only concern yourself with the application itself. Scalping to zero isn't severless, but a consumption model. The consumption model is really only possible within a serverless environment, but is not synonymous. I think the issue here is likely less with what serverless means, as that seems self evident by the name, and more with a associating it with something that it didn't mean. That doesn't mean that association wasn't reasonable at the time, but just there was a misunderstanding and nothing has really changed.
Price of opensearch is insane
I think serverless just means you don't need to or even get to worry about the server. Like applying system patches. Or worrying about mount points. Our whatever server things you don't think about anymore. That doesn't mean it doesn't actually run on a server, or that it scales to zero
I get your definition and the hopeful thinking, but what they mean by serverless is "(mostly) aws managed resources allocation".
The price is driven by the service where AWS takes away the complexity, not how fancy the compute/storage is.
It's their world we work in and that's their created term.
It has nothing to do with what the customer wants, it's all about what the tech giant knows the customer will pay for.
The problem with those services is the minimum level is often too high.
True serverless relational DBs are still something that still don’t exist.
Stop using proprietary technology, take servers and run open source. Have control over what you get.
Hi Daniel 👋 new sub here!
My website users upload 3 to 6 json files. I'm considering to save these files in a s3 bucket. Should I save the links to json files in a database like mysql?
Your tutorials are so easy to follow, please keep the great content rolling 🎉
If the json files follow a predefined schema, the json files could themselves be in the SQL db. Perhaps a different table linked via a key.
You will retrieve info faster, and write less code.
I appreciate your sentiment and completely agree that scaling to zero ought to be an essential component of the definition of serverless.
This couples particularly poorly with AWS' frustratingly non self-describing branding of their services, especially compared to its Azure and GCP counterparts.
I think it would be prudent to iron out their branding overall, making sure serverless means the same across the board and maybe doing something overall with their service/resource naming convention to append something or alter something to make their naming make sense.
Thanks
It’s really bad manner to offer virtual currencies in order to mask real costs.
amazon is a big face palm if you talk about database serverless
serverless for the consumers but not for the vendor (amazon) 😀
They are managed services, not serverless..
anything but AWS Lambda is the only serverless
Turns out my 'unlimited' data plan is actually quite finite. I don't care what a vendor calls something these days, show me the spec sheets.
And it has really bad performance compared to raw cheap ec2 instances.They are scaling for themselves not for customers.
Managed service is more accurate
I never really understood the buzzword "Serverless" I mean there really isn't a such thing as being severless when using the cloud. I mean what's a Cloud platform without the physical underlying infrastructure that runs it? Just because you may not manage it, doesn't mean you aren't using physical hardware that runs thr Middleware and everything regardless if it's PaaS or IaaS. You are still using some else's hardware infrastructure.
Serverless just means you do not need to concern yourself with the hardware or OS. You only concern yourself with the application you are running, and let AWS, Azure, etc. handle maintaining the hardware.
@@evancombs5159 That's more like application deployment but again, all of that is STILL running on a physical server infrastructure in a data center regardless if you don't manage it. No different than accessing a website on the web. You are still interacting with a server. The client sends a request to the server and server responds and exchanges the information back.
I don't see how something like Aurora could be 'server-less'.
Wow
"2024 is the year of serverlesslessness" - Programmers are also human
This is a total piss-take. I agree, Serverless, these things are NOT, and the naming needs changed to relect that. Really angry 😡that this is actually a thing, and it needs nipped in the bud -- it has somehow become "normalised".
Even the original concept and name "serverless" was a bit misleading. It sounds like magic. How does the code run without a server?? Server-on-demand might have been a better name.
At reinvent they had a talk about server-less best practices and in the first 5 minutes touch on what sever-less was 10 years ago and what it is now… ua-cam.com/video/sdCA0Y7QDrM/v-deo.htmlsi=vZHh0d-srZyGR2pV
What ARE you talking about? Serverless is NOT scale-to-zero! People got that idea stuck in their head when lambda came out and worked that way! Serverless means exactly one thing, and _ONE_ thing _ONLY:_ YOU don't rent a server, neither directly, nor as a consequence of running an instance of a service, an RDS database for example.
And about the cost: You have to keep an eye on your cost and calculate them ahead of time when running services in the cloud? Shocking development, Sherlock!
Seriously, the REAL problem is that with AWS' low-bar-services like Amplify, you get a lot of devs on the cloud who are in WAY over their heads. You need a license to drive a car, but you don't need a license to mortgage your income for the next twenty years in the cloud, because you couldn't be bothered to do the math ahead of time.
Maybe what AWS should really implement is a kiddie mode that just pulls the plug at a certain monthly cost and to hell with anything but backups.
Azure has that feature, you set a cost limit and things turn off once that budget is met. It is also my understanding that Azure is generally cheaper than AWS so a lot of this may be less of an issue over there if my understanding is correct.
The term "serverless" doesn't even make sense anyway
Serverless or not, when you buy into all this cloud nonsense, don't be surprised when you discover that you are being ripped off by and completely dependent on a big company. Why would anyone "run their reports on Friday" on AWS if this can be done in-house by building your own server? Self-host everything that you need and you will never have to spend time and energy parsing through yet another portion of big tech marketing garbage. And if you can't self-host something, you should not do it at all.
The big reasons are uptime and scalability. Two of the hardest things to do with on-prem is make sure there it's not downtime due to network or server issues, and to quickly scale a server as needed (i.e. have normally unused server capacity). There are other things like location redundancy. These are things even larger companies struggle with, let alone small or medium sized companies.
There is so much more wrong with AWS. It is very hard to find actual specifications, just propaganda everywhere. It is waste of time. Just hire proper sys admin and have your dedicated server running at fraction of cost.
What are these comments talking about? By the logic GKE is serverless. What a bunch of rubbish takes.
whats up