Introduction to Cloud Computing
What is the Cloud?
means storing data and accessing computers over the internet
it is not just storage, it'd compute, it's artificial intelligence, machine learning, etc.
it's a bunch of different types of computing or services, and just being done on somebody else's server
What are the Origins of the Could?
'60s to the original ARPANET project when they were building and architecting the internet
cloud symbol was used on their flowcharts
the idea being there that any computer that's connected to the internet has access to that pool of computing power and data
late 1980s or early 1990s, the way people would install software on their computer was using floppy disks
a good analogy here is like electricity for cloud computing
you can generate your own electricity
you could buy a generator
you could run it in your house
relatively expensive investment up front
got to buy that hardware
got to keep it running
got ongoing maintenance of your generator
if there are enough people around you all wanting access to electricity, it kind of makes sense to share that generation with everyone else
outsource the setup and running of the generation of electricity to someone else like the electric company
then anyone who needs electricity around you can just tap into it on demand, and they can just pay for what they use
it allows the energy company to focus on delivering energy, producing energy in a really efficient, fast way and build economies of scale
it allows the consumer of the energy to focus on consuming the energy to do what they want to do instead
Who are the Different Cloud Vendors?
Amazon, by far the biggest
make up around 90% of the public cloud compute
followed closely by Microsoft Azure
actually make up the next 5%
final 5% roughly is Google, IBM, Alibaba, etc
Google is exceptional at big data analysis and machine learning, deep learning, and artificial intelligence
How Did the Cloud Come About?
early days of the internet
if you wanted a server, you needed to run it yourself. You needed to buy one, you needed to set it up, power it up, plug it in, connect it to the internet, and you needed to keep it running, 24/7, keep your internet connection on
people realized they could outsource this
data centers came along in a huge way
you could go to a data center, and they would set up racks of servers that you could rent for them
at first, they provided only infrastructure support, so they basically would provide those servers with power, with heating or cooling, electricity, that sort of thing, but then later on, people were requesting for Rackspace to support the operating systems, or to support the application layer
still large upfront investments
pay for those servers, you had to sign lengthy contracts with the data centers
about 2006 and Amazon comes out and launches their first cloud platform, they launched EC2
the benefit of it was that you didn't have to worry about providing electricity, you didn't have to worry about providing the internet, they managed all that for you, the physical security
the journey to cloud has really been occurring over from '95 through to about 2006
What's a Server?
just like a powerful computer that you're able to access
typically rack-mounted or put into blade chassis
won't have a monitor that you can connect to
connect into them remotely
What is Serverless?
the cloud provider like Amazon manages all the infrastructure for you
with serverless is give my code to Amazon or to Microsoft or to Google
run it whenever a customer needs something
they deal with the challenge of scaling that
they deal with all the operations of having people available 24/7 to fix things when things break
they provide guarantees to me as a customer, SLAs, and I can trust them to run my code
What is LAAS?
Infrastructure as a Service (IAAS) is where you have a hypervisor, and the hypervisor you can provision virtual machines
it could also be where you're actually renting the physical server itself
essentially it allows you to host an operating system in the cloud, which then is unique to you
nobody else can log into it, and you go in and then manage it yourself
you only pay for what you use, so it depends on the model and the idea is there's no long-term or fixed contract
Amazon:
it's a per second billing model right now for Linux instances
windows instances are on a per-hour basis
Google:
a per minute basis
What is PAAS?
Platform as a Service (PAAS) is basically something like Elastic Beanstalk
don't necessarily know what resources you need, but you've just got your code, and you will then use PAAS to go in and provision those resources for you
Elastic Beanstalk will provision you with elastic load balances, with EC2 instances, with databases, etc.
you still have to look after the underlying assets, but you don't have to worry about the provisioning of it
What is SAAS?
Software as a Service (SAAS) is something like Gmail
With Gmail, all you worry about is the actual software
worry about your inbox
worry about spam filters
not worried about the underlying servers
don't have to worry about how they load balance it
don't have to worry about how they provide high availability
don't have to worry how they do their DNS reading
What is FAAS?
Functions as a Service (FAAS) is somebody else goes out and writes a function, which you can then use
you can break down your application from these monolithic applications into something that's a lot smaller and decoupled
Ex:
somebody has written a function where you just take a payment
you would then use that code and recycle that code in your own application
What Advantages Are There Using the Cloud?
things like big data and analytics require an awful lot of computed power and also an awful lot of servers, especially if you're scaling out horizontally
if you were to do this yourself, you'd have to go out and buy these servers, you'd have to rack and stack them, power them up, and then you'd have to go through and manage them
if you're buying from HP or Dell, you're in three- to five-year contracts, so it's going to be very, very expensive
you pay by the minute or by the hour, depending on the service, and you can go in, do your big data calculations or analytics, and as soon as you're finished, you can stop
not locked in to a three- to five-year contract