We use Amazon S3 Glacier for hosting web files for testing.
We take a lot of data from the police to do ANPR. It's basically for big data to allow bandwidth on the data center, where there are about 1,000 images per second coming through to the data center.
AWS Lambda is the most valuable part of the solution because it gives you a code interface for everything.
Their pricing structure is always a big topic. It's so difficult to understand. I had a personal account once, and I was so scared to type in my credit card because I wasn't sure how much I would be charged.
At home, I might turn on the VM to test something, and I might be on there for three minutes wondering how much it could cost. The solution might tell you there are so many hours of free usage and a free VM, but it's still scary because you don't know how much it will cost, even if you go over one second.
The price structure needs massive simplifying and made crystal clear so people know what they're paying.
I've used Amazon S3 Glacier for years.
I rate the stability a nine-point nine out of ten. They've got shared data services, giving you an SLA of about 0.001%.
There's a huge amount of data on Amazon's data center. Every camera in the West Midlands takes photographs of cars constantly and loads all the data into the data center. Since Lambda is Python, I'd rate the scalability a ten out of ten. I understand the API Gateway completely. But when it comes to security, I'd rate it a three out of ten. AWS, as a whole package for what it provides, I rate its scalability a seven out of ten.
The technical support team has a lot of people who know their stuff.
Positive
The initial setup is difficult, not because of applications like S3 or VMs or firewalls and gateways, but because of security. The most important thing about using AWS isn't the tools you will use. It's the security you apply to it. The solution's security is a complex area involving services that talk to each other and how the users interact with those services, whether internal or external, to be logged into AWS. The people in security need to be certified.
The deployment was a continual process, working with several external organizations, such as the transport police in my case. Setting up Amazon takes under an hour if you've got everything available. But the process starts with knowing the tools we need to achieve what we're trying to do. Amazon is very good at talking, explaining what they've got, and upselling. They'll tell me exactly what you need, though they won't tell me how much it costs. But that process takes months to understand unless we get free services to trial and test the solution before you start going into it.
I've seen value and impact in my organization using Amazon S3 Glacier.
You need a PhD to understand their pricing model. We often sit through meetings, going through all of these documents from AWS, going through how to calculate pricing based on so many different factors. At the end of it, there's a fear when the bill comes in because nobody understands how much it's gonna be.
AWS has these pricing calculators. You type in which server you want and how much usage, but at the end, you don't feel confident when it says $3.98 because I know that we were paying £20,000 per month for AWS services at one point. Some of the other services are RedShift and something else. When pulling in data for the police for ANPR, storing the data in the UK data center costs a fortune.
I rate Amazon S3 Glacier a ten out of ten.
Two bits of advice: make sure you get a specialist with things like security, such as network security. The security is indirect with AWS. You could get an expert in Python using Lambda, but it doesn't make you a network security person. Once you've got a million pounds worth of business data in Amazon, you could potentially share that with the planet if you don't know how to set up the security. Get the relevant technical specialist for things like security.
The second bit of advice is to completely understand the pricing structure.
