Managing Seasonal Traffic with AWS

The ROI could be a big blocker when it comes to businesses. Infrastructure cost often is the most expensive for running an application. Businesses usually procure on-premise hardware to scale or buy online virtual machines as the business grows but is it really scalable?

Let see an example for a business having on-premise infrastructure & consider an online freedom sale day where:

• Usual traffic for a business is let’s say 500,000 users a day.
• To handle such traffic let’s consider 100 web servers are needed and 500 MBs per second bandwidth too. It could be on-premise set up or even buying compute on a cloud service provider platform.
• Its Diwali and the traffic is predicted to increase to 1,000,000 users a day, which means at least 40% new servers need to be procured.
• Assuming the Diwali campaign went well, post which traffic from 1,000,000 users a day decreased to let’s say 300,000 users. What does all this mean to the business?

1. Sudden increase in costs
2. It is a time-consuming activity to procure and setup these servers
3. Loads of management & maintenance overhead
4. Need to forecast traffic well in advance
5. Hardware once purchased can’t be returned

Enters, Amazon Web Services.

Any resource you use on AWS cloud follows an on-demand model out of the box. What does that even mean?

Let’s consider a similar example as above where a business has :

• Traffic per day – 500,000 users (HTTP requests)
• Currently, traffic is being handled by 100 AWS EC2 instances acting as web servers

What will it take to scale to handle 1,000,0000 users?

The below:

1. One-Click increase in compute capacity to as many servers as you need
2. You are charged only for the number of hours you use these servers for
3. Reduce the capacity back to what it was, only with a click of a button
4. High & best in industry bandwidth is a standard on the Platform

No management overhead, no procurement procedures.

You pay only for what you use and how long you use it.

Now, that you’re interested in knowing more, let’s see how AWS can provide you with more scalability and flexibility.

1. Autoscaling
Most of the times we know but sometimes we have no idea about how the user traffic for our applications would be distributed during the course of the day.

During some hours the traffic might be really high as compared to the other.

AWS Autoscaling helps scale the count of instances based on various thresholds like traffic, compute utilization etc. What that translates to is that one decides the number of instances based on usual traffic they see. Once that is done, thresholds can be set for various metrics and if that threshold is crossed, autoscaling will itself scale up instances to meet traffic demand.

As soon as the traffic is normal, the instances will be scaled back to default. During very low traffic hours, probably at night, autoscaling can help save huge bucks by downscaling instances during those hours of the day. Scale things back down after the busy times.

2. Spot instances
Spot instances are instances available at rates cheaper than normal EC2 instances.
AWS offers the unused capacity that it has to be bought by users at a cheaper rate. Though it comes with a trade-of where AWS might take the instance away from the user anytime with a 2 minutes warning, the cost saving it has to offer is just mind blowing.

For example:
On-demand 2 core, 2GB Instance cost: $0.01 per hour

Spot instance of same configuration cost: $0.005 per hour

3. Know your costs upfront
AWS S3 price calculator & Total Cost of Ownership calculators help users know what the cost of various services on AWS would be for different usages and calculate what savings can one expect moving into AWS cloud respectively. What this means is even before you move into AWS Cloud you can know what your costs would look like; whether on-premise or cloud solution suits your use-case.

4. Always available compute capacity
Compute on the AWS platform is infinitely scalable. One can procure as much compute or storage capacity as they need. AWS always has tons of extra capacity available for users at any point of time.

5. Serverless computing
AWS also offers serverless computing. No need to worry about infrastructure management anymore and in AWS words “increase your agility and innovation”.
It also offers fault tolerance, availability, scalability & ZERO management out of the box. No need to keep anything up and running all the time.

For example:
You need to send users order receipts. Usually you would have another set of servers just to handle this microservice. Other way is you could simply put this code on AWS Lambda, and let it take care of the use-case.
One could trigger the code millions of times and still spend very less and also not pay any hourly or daily costs.

AWS even offers serverless databases like Amazon Aurora Serverless reducing your overhead of managing any sort of infrastructure yet having access to a full fledged database.

6. Move workloads into AWS cloud easily
If you already have an on-premise setup or you’re based out of different cloud platform provider, you could still move your workloads into AWS Cloud with utmost ease and make use of this scalable platform for any seasonal traffic.
One can still continue to use the tech stack they wish to and easily plug AWS cloud into it.
AWS offers loads of migration services like AWS Server Migration Service etc. which facilitate the same.

To get the most of these alluring AWS services and more, get in touch with BlazeClan.