Optimizing AWS cost
By default, we deploy the ‘light’ version of the Cloud VMS that uses t3a.small EC2 instances that are much cheaper and are suitable for the small load (less than 100 cameras). This approach significantly decreases the idle cost of the system in the first weeks of onboarding cameras. If the usage increases, the type of instances is upgraded to m5.large which is suitable for production.
There are several other ways of optimizing AWS cost:
1. Stop instances that are not used
The VMS is a group of microservices and some of them are hosted on separate instances. So if a service is not used, it is a good idea to stop the corresponding instance. For example, if your use case doesn’t require live image generation, the LiveImageGen instance can be stopped (stopped by default) or if you don’t need low latency streaming and HLS is sufficient, then the WebRTC instance can be stopped.
2. Use reserved AWS instances
AWS provides a significant discount (up to 40%) if you reserve instances for 1 or 3 years instead of using by-demand instances. Please contact Amazon for more information.
Read more:
3. Apply for AWS credits
AWS provides credit packages (up to $100,000) that can be used to cover AWS bills to help startups reduce costs and increase speed to market.
Read more:
4. Use a dedicated network connection
This option can work better for monitoring centers when you need to watch/download many videos.
The standard egress cost is $0.09 per GB (down to $0.05 with volumes).
The cost of transfer through a dedicated connection is $0.02 per GB, plus there will be a cost for network port hours ($0.30/hour for 1Gbps).
The total cost for a monitoring center watching 100 channels simultaneously will be the following:
Data - 36 TB/month (720 hours at 1Mbps, for 100 channels), port hours - 720 hours
Dedicated connection | Standard egress |
|---|---|
$720 + $216 = $936 | $3,240 |