The Cost of Serverless

In this video, I talk about the cost of serverless
links from video:
/ 1011871609039171587
haveibeenpwned...
www.digitaloce...
aws.amazon.com...
----
Join the Discord: / discord
----
Patreon: / benawad
----
Follow Me Online Here:
GitHub: github.com/ben...
LinkedIn: / benawad
Instagram: / benawad97
Twitter: / benawad
#benawad
TikTok: / benawad ​
​
----
Follow me online: voidpet.com/be...
#benawad

[@MarkDeibert]

I use and love Digital Ocean but I'm also super facinated by serverless setups. Please keep this video series going! As always, excellent channel Ben :-)

[@LetterSignedBy51SpiesWasA-Coup]

i like your approach of controling as much of your stack as possible and deploying in containers. serverless and SaaS gets very vendor-specific. also debugging is a much better experience when the stack is running on your machine. for instance with AWS, i recently came across a situation where only successful calls were logged in Cloudwatch and I had to guess to solve the problem because no errors were being returned in the response.

[@bawad]

yeah, debugging in AWS can sometimes be a nightmare

[@maarten1012TTT]

Most of the time you can use localstack though

[@dr.z7958]

AWS, Azure, GCP are definitely not cheap. Especially for bandwidth (it's insanely expensive, in most cases you'll find +30% of your cost is from bandwidth). However, developing cloud-native app, gives an unlimited potential of designing your services, and then re-design them in very short time (in case you grew and noticed some solutions cannot keep up with such growth). The second thing, is aws, azure, gcp is being cheaper every year (due to strong competition).

[@lardosian]

Have a new role where we will be building an API and a PWA for a startup, the startup has AWS credits so I've been doing research and think the serverless framework is a good way to go, along with api gateway, lamda functions, dynamodb, cognito and S3. Any advice would be great, thanks!

[@zakmorgan9320]

Handling that many requests is fairly trivial, like others said the lowest droplet tier should accomplish it, I've handled similar requests for research projects handling real-time data from participants recently. In terms of how digital ocean can offer the bandwidth so cheap is a combination of two things that's typical for VPS providers. First is the average user on digital ocean probably uses very little traffic. The second is that if you actually use that full 1TB a month I guarantee you'll find your account gets terminated for breaking their terms of service. Usually if you actually used the full resources of a VPS it's considered a violation of their fair use policy.

[@danielfarhat799]

Interesting. Is there a reference to the fair use policy you mentioned ?

[@rtorcato]

What's he doing with the blob storage? maybe his images are stored in a database instead he could've put on s3. With that much traffic he is laughing at that small bill. i know companies spending more on servers with a fraction of that traffic.

[@bawad]

Maybe he stored all the passwords in there 🤷

[@thehumanchannel9004]

The question is does Digital Ocean provide Serverless or do they have any service that charges websites based only on what they use

[@kifkev91]

For the question of what hardware you need, couldn't you just resolve that by writing a script that hits your webapp on your own machine as fast as possible. For example, if you built a blog , you write a script that creates 10.000 entries. If you measure the time, how long it takes to execute, you can calculate how many requests your app can handle per second with one cpu core. Then you can cluster your app for how many cpu cores your machine has, and measure how your machine handles that. I don't know if my thoughts are flawed here, but to me it seems quite easy

[@bawad]

Yeah I was thinking of trying something like that