Let’s start with the basics. What exactly is multi cloud? At CAST AI, we believe that real multi cloud is when you have the freedom to create, deploy and move containerized applications across several clouds, and when you can stretch applications across several clouds simultaneously. In 2020, 93 percent of enterprises already have a multi cloud strategy (Flexera).
However, going multi cloud presents a challenge: It requires compute and other cloud-native resources from at least 2 cloud providers to be located “nearby” from the perspective of latency. For instance, 5ms and 10ms become natural barriers to federating resources into one pool. Beyond these figures, applications start to break.
To benefit from multi-cloud, organizations need to identify geographically close zones that have an offering of several cloud providers. We call these areas multi-cloud Goldilocks Zones. In these zones, businesses can achieve the best results in balancing cloud cost and performance.
At CAST AI, we have found many Goldilocks Zones where multi cloud deployments can thrive. Read on to find out how we did it and how you could benefit from multi cloud in Goldilocks Zones too.
What is a Goldilocks Zone?
The term “Goldilocks Zone” comes from astronomy/astrobiology and describes places that offer a balance of conditions required to support life. In our context, the Goldilocks Zone refers to a situation in which organizations have the right resources and components to support the “life” of their applications. Balance is a key concept associated with Goldilocks Zones. The name was taken from a British fairy tale in which a little girl chooses from a set of three items, ignoring ones that are too extreme and settling for the one in the middle, one that is just right.
In the context of multi cloud, Goldilocks Zones refer to geographical areas where multi cloud is made viable and thriving. A Goldilocks Zone is a location:
- where at least two different cloud providers are available,
- and where there’s stable transit between each of the cloud providers with sub 5ms, 8ms, or 10ms maximum latency.
Here’s an example of a multi cloud Goldilocks Zone: EU Frankfurt, 5 millisecond latency.
Why should I care?
Why is finding a Goldilocks Zone that enables multi cloud important for your business?
1. Cost optimization: It offers opportunities for cost optimization across different cloud providers, benefitting from the most cost-effective offers available in a given region.
For example, these two instances of 4 virtual cores, 16GB. Google is 10% less expensive than AWS, and offers 7% more compute capacity in EU Central Frankfurt. Hence in a perfect world, Google Cloud is more cost effective and should be selected over AWS, for a 4 core VMs in EU Central Frankfurt.
2. Performance optimization: You can optimize performance across various cloud platforms. For example, if your application experiences a sudden increase of requests per second, your software developers can quickly check which cloud platform within the Goldilocks Zone has the available resources at the lowest possible cost and start deploying additional capacity immediately.
3. Disaster recovery… or simply adding capacity: If a cloud provider within the region goes down, you can quickly shift your workloads to other cloud platforms available in this region.
To sum up: By deploying a containerized application in a multi cloud Goldilocks Zone, you can stretch it across various cloud providers without risking any loss of performance thanks to the robust network capabilities, with all the benefits from multi-cloud, from cost optimization and disaster recovery to capacity management.
Here’s what we found out about cloud Goldilocks Zones
At CAST AI, we’ve had our eye on potential Goldilocks Zones for a while now and learned a lot while searching for them. Here’s a summary of our process:
- You can find a Goldilocks Zone with almost all the top providers in all the relevant markets.
- We managed to identify many Goldilocks Zones where services from at least three major cloud providers were available.
- We compared all of the resources available in these Goldilocks Zones to check which ones were the best fit for a multi cloud approach.
- We now know where going multi-cloud is possible, and we can provide the information about all of the available resources and the price/performance ratio in these zones.
At 5ms, there are 18 clusters of Goldilocks Zones that include 57 regions between AWS, Azure, Google Cloud, Digital Ocean and Oracle Cloud.
Below is the one for Singapore:
1 |
Asia Pacific (Singapore) – Amazon Web Services |
2 |
Southeast Asia – Microsoft Azure |
3 |
Jurong West, Singapore – Google Cloud Platform |
4 |
Singapore 1 – DigitalOcean |
At 8ms, there are 14 larger clusters Goldilocks Zones that include 65 regions. Some of these zones are Mega Cloud Regions; zones that contain a huge number of cloud providers within 8ms. For example, we found 21 cloud platforms forming the Mega Region of EU Central:
1 |
Switzerland North (Zurich) – Oracle Infrastructure |
2 |
St. Ghislain, Belgium – Google Cloud Platform |
3 |
West Europe – Microsoft Azure |
4 |
London, England, UK – Google Cloud Platform |
5 |
Frankfurt, Germany – Google Cloud Platform |
6 |
Eemshaven, Netherlands – Google Cloud Platform |
7 |
Germany West Central – Microsoft Azure |
8 |
France Central – Microsoft Azure |
9 |
London 1 – DigitalOcean |
10 |
Europe (Paris) – Amazon Web Services |
11 |
Switzerland North – Microsoft Azure |
12 |
UK South (London) – Oracle Infrastructure |
13 |
UK South – Microsoft Azure |
14 |
UK West – Microsoft Azure |
15 |
Frankfurt 1 – DigitalOcean |
16 |
Europe (London) – Amazon Web Services |
17 |
Amsterdam 3 – DigitalOcean |
18 |
Zürich, Switzerland – Google Cloud Platform |
19 |
Netherlands Northwest (Amsterdam) – Oracle Infrastructure |
20 |
Germany Central (Frankfurt) – Oracle Infrastructure |
21 |
Europe (Frankfurt) – Amazon Web Services |
At 10ms, more mega regions are forming, including the Mega Region of US East:
1 |
East US – Microsoft Azure |
2 |
Canada Central – Microsoft Azure |
3 |
Canada Southeast (Toronto) – Oracle Infrastruc… |
4 |
Canada Southeast (Montreal) – Oracle Infrastru… |
5 |
Ashburn, Northern Virginia, USA – Google Cloud… |
6 |
New York 1 – DigitalOcean |
7 |
US East (Ashburn) – Oracle Infrastructure |
8 |
Canada (Central) – Amazon Web Services |
9 |
Toronto 1 – DigitalOcean |
10 |
Montréal, Québec, Canada – Google Cloud Platform |
11 |
New York 3 – DigitalOcean |
12 |
US East (N. Virginia) – Amazon Web Services |
At 15ms, all regions (save for 4 out of 84) are in a Goldilocks Zone inside 13 Mega Regions.
So… How did we do it?
CAST AI has active resources (VMs) in 84 regions and is continually measuring traffic and transit performance between all the regions [84 x 84], in total 3528 network pairs. Our data science team has built an AI engine that assembles Goldilocks Zones based on clustering regions for the traffic and stability performance into four groups: 5ms and less, 8ms, 10ms, and 15ms.
Create your first cluster and try it yourself: {{cta(’51a429e5-c069-419c-8aff-81cd7161c294′)}}
Now you can do it too!
To avoid vendor lock-in and implement the most suitable solutions, it’s smart to use multiple cloud services for deploying your applications. To make multi cloud possible, you first need to compare your options. Read this article to discover the easiest way to do it: How to compare the cost of AWS, Azure, Google, and Oracle, once and for all. The second step is finding out whether multi cloud is possible in a given region, considering the network performance.
Follow this link and find out whether you’re overpaying for your cloud resources.