Fundamentals - Core Infrastructure Flashcards
What is Cloud Computing?
Cloud computing is a way of using IT which has 5 traits:
- On-demand self-service: No human intervention to get resources
- Broad network access: Access resources over the net from anywhere
- Resource pooling: Provider has a big pool of resources they share with customers (economies of scale get passed on from provider to customers)
- Rapid elasticity: Get more resources quickly as needed
- Measured service: Pay only for what you consume or reserve
What does IaaS stand for?
Infrastructure as a Service
What does PaaS stand for?
Platform as a Service
What does SaaS stand for?
Software as a Service
How are Google Cloud Platform resources organised?
By zones (e.g.: europe-west-2a, europe-west-2b, europe-west-2c) within regions (e.g.: europe-west2)
Is a GCP zone the same as a datacentre?
No - a given zone can be distributed across multiple buildings.
Why would you want your resources to be located on zones within the same region?
All the zones within a region have fast network connectivity between them
Distributing on multiple zones may give you more resiliency
What is the average round trip latency between zones within the same region?
Under 5ms
Why would you want your resources to be distributed on more than one zone?
To make your application more resilient/fault tolerant.
It helps protect against un-expected failure.
Can you run your application on more than one GCP region?
Yes
Why would you want to configure your application to run on more than one region?
Latency: creating resources close to the customers you want to serve reduces latency.
Resiliency: e.g.: in case a natural disaster takes region down.
Do all GCP resources have to be provisioned to a specific zone?
No - some GCP resources support being configured as “multiregion” (e.g.: Google Cloud Storage)
What does it mean if a GCP resource has been configured to a given multiregion?
It means the resource is stored redundantly
in at least two geographic locations …
…separated by at least 160 kilometers
Roughly how many regions does GCP have?
15
Name GCP’s Infrastructure as a Service (IaaS) offering
Google Compute Engine (IaaS)
Is per-second billing available on GCPs Infrastructure as a Service (IaaS) offering available?
Yes - Google was the first to implement per-second billing on it’s IaaS offering (Google Compute Engine)
What type of workloads might benefit from fine-grained billing?
Ones which come in bursts as opposed to steady constant streams: only pay for the time you use rather than a whole hour.
Name 2 GCP services that support per-second billing
Google Compute Engine
Google Kubernetes Engine
Which GCP service offers “sustained use discounts”?
Google Compute Engine
What is a “sustained use discounts”?
An automatic discount
that you get for running a virtual machine instance
for more than 25% of the billing month
How long do you need to be running your virtual machines during a given billing month to benefit from the sustained use discounts?
More than 25% of the month
How can you estimate your costs before setting up your application in GCP?
GCP has an online calculator to help you estimate your costs.
What makes GCP pricing customer friendly?
Supports:
- Billing in sub-hour increments
- Customer VM types (only pay for the resources your workloads need)
- Discounts for:
- Sustained use (when VMs are used for > 25% of the month)
- Committed use (pay less for steady long-term workloads)
- Preemptible use (pay less for interuptable workloads)
What is a preemptible workload?
A workload that can be interrupted mid run.
Why choose Google Cloud Platform?
It enables developers to: * Build * Test * Deploy On Google's infrastructure which is: * Highly secure * Reliable * Scalable
Google Cloud Platform’s products and services can be broadly grouped into 6 categories; name them.
- Compute,
- Storage,
- Big data,
- Machine Learning,
- Networking,
- and Operations and tools
What tools does GCP provide to help protect you from running up an unexpected big bill?
GCP provides four tools to help:
- Budgets and alerts,
- Billing export,
- Reports,
- and quotas.
What can you define GCP budgets against?
Per Billing account
Or per GCP Project
What are the 2 different ways you can define your GCP budget limit as?
You can set your budget limit as:
- A fixed amount
- Or tied to another metric (e.g.: a percentage of of the previous month’s spend)
What type of value do you use to define your budget alert triggers?
A % (e.g.: alert me when I’ve spent 90% of my budget)
What are the preset values for GCP budget alerts?
50%
90%
100%
But these values can be customised.
Why would you use GCP Billing export?
Billing export lets you store detailed billing information in places where it’s easy to retrieve for more detailed analysis,
Where might you export your GCP Billing data to so that you can analyse your spend?
You could export it to a Big Query dataset
Or to a Cloud storage bucket
What is GCP Reports?
GCP Reports is a visual tool that allows you to monitor your expenditure
Why would you set-up GCP quotas?
To protect both account owners and the GCP community as a whole
by preventing the overconsumption of resources,
whether because of error or malicious attack
What are the 2 types of quotas you can define in GCP?
- Rate quota
* Allocation quota
At what level do quotas get defined (e.g.: resource, folder, project, organisation)?
At the GCP Project level
What do rate quotas define/limit?
The number of calls per period of time.
The rate quota is reset after the period of time has elapsed.
For example, by default, the Kubernetes Engine service sets a quota of 1,000 calls to its API from each GCP project every 100 seconds.
After that 100 seconds, the limit is reset.
What do allocation quotas define/limit?
They limit the number of resources you can have in your project
For example, by default, each GCP project has a quota, allowing it no more than five virtual private cloud networks.