Microservices or Client-server Model

[mcmiller]

I recruit in IT, and my team is working on a web-app, but we are getting conflicting information. Our web-app should get 10s of thousands of daily traffic. We are concerned about scaling up a client-server model. We think microservices or "serverless" model might be the solution.

 

We we speak with the companies that will help us with this, all we hear is bias and self-interest. Do any of y'all have an experienced opinion on the matter? Pros and cons... what should we look out for?

Edited June 16, 2020 by mcmiller
spelling mistake

[Ghideon]

13 minutes ago, mcmiller said:

We we speak with the companies that will help us with this, all we hear is bias and self-interest. Do any of y'all have an experienced opinion on the matter? Pros and cons... what should we look out for?

With the limited information available my answer will be very general.

I would be cautious if someone provides detailed answers based on limited information.

10s of thousands can be a tiny or huge depending on the amount of data, how the traffic is clustered, response time requirements, processing required to create an answer, what standard services that are already available as services, API, micro services etc. 

 

Edited June 16, 2020 by Ghideon

[mcmiller]

10 minutes ago, Ghideon said:

With the limited information available my answer will be very general.

I would be cautious if someone provides detailed answers based on limited information.

10s of thousands can be a tiny or huge depending on the amount of data, how the traffic is clustered, response time requirements, processing required to create an answer, what standard services that are already available as services, API, micro services etc. 

 

Yeah, it's tough because I have no idea how to formulate a conversation on the topic. Basically, it's a dynamic web app, there will be .doc and .pdf documents uploaded and viewed... and some text boxes... practically no photos, NO videos, etc.

[fiveworlds]

Quote

Yeah, it's tough because I have no idea how to formulate a conversation on the topic. 

I assume you aren't making the final decision so don't take that responsibility. You should create a PowerPoint presentation referencing the various options that are available and present that to the other members of your team so an informed decision can be made. If you are uncomfortable doing so yourself I'm sure you can recruit a consultant on a temporary basis to do so.

 

[Sensei]

2 hours ago, mcmiller said:

Our web-app should get 10s of thousands of daily traffic.

Default Apache HTTP server config allows 256 concurrent requests. That's 22 millions per day..

2 hours ago, mcmiller said:

Basically, it's a dynamic web app, there will be .doc and .pdf documents uploaded and viewed... and some text boxes... practically no photos, NO videos, etc.

Sounds pretty light-weight.. Unless the all people will start using service at the same hour..

 

Average size of document?

Estimate amount of data transfer per day.

Client = user web browser from computer? or mobile application written for Android/iOS?

Server = your computer? Virtual server that you rent from 3rd party? Dedicated server that you rent from 3rd party?

 

If client = user with web browser, and server is your own machine, the easiest is to buy cloud and store data on it.

[Ghideon]

3 hours ago, mcmiller said:

Yeah, it's tough because I have no idea how to formulate a conversation on the topic.

Ok. 

3 hours ago, mcmiller said:

Basically, it's a dynamic web app, there will be .doc and .pdf documents uploaded and viewed... and some text boxes... practically no photos, NO videos, etc.

Ok. That's not much to base an analysis upon, other than that it sounds like a description of a lot of existing services. What makes the application worth creating instead of using an already existing application? Why will users use this application instead of any other? The answer to that will give some important hints about how to approach questions about suitable architectures.