Follow Mark on LinkedIn Follow @marknca on Twitter Follow marknca on YouTube

imgs/hero.jpg

Cloud Madness

13 minute read |An icon depicting a retail tag with a heart for 'favourite'cloud

Every March brings the best of college basketball in the United States together to compete for a national championship. March Madness is an exciting series of tournaments that leads to 68 men’s and 64 women’s teams facing off to see who’s the best.

With a nod to that tournament and how it shows off all of the different ways to play the game, Forrest Brazeal and the team at A Cloud Guru have put together “Cloud Madness”.

32 cloud services from Microsoft Azure, Google Cloud, and AWS face off. the Gurus will be running various polls to determine the winner of each match up throughout the month. But, as is Madness tradition, I wanted to write out my bracket before the tournament started.

Here’s the thought process behind my picks.

Round #1

North West / Top Left

Amazon EC2 vs Azure Virtual Machines

Both of these services are fantastic. If you need a full operating system, it’s hard to go wrong with either of these. I give the edge to EC2 here. EC2 is the closest thing the cloud has to a dynasty. This service has stayed at the top by continually innovating and adapting to the competition.

Winner: Amazon EC2

Community Winner: Amazon EC2

AWS Lambda vs AWS Fargate

One of the toughest match-ups in the entire tournament. I have to give the edge of Lambda here because it’s the OG of serverless compute. While Fargate has come a long way since it’s—overdue—launch. It just can’t overcome the agility of those Lambda powered serverless applications.

Winner: AWS Lambda

Community Winner: AWS Lambda

Google Cloud Run vs Amazon ECS

In a runaway win, Google takes this one. This service offers a huge amount of serverless value to containers with only minor modifications. It’s a pragmatic service that accepts that sometimes you have to work with what you’ve got. Watch Bret McGowen’s talk from ServerlessConf NYC 2019, if you’re new to this service.

Winner: Google Cloud Run

Community Winner: Amazon ECS

Google Kubernetes Engine (GKE) vs Azure Kubernetes Service (AKS)

Again, a strong match up but GKE has a much longer history (both public and private). If Kubernetes is part of your tech stack (and you seriously consider if it should be), it’s critical infrastructure. Stability and track record wins out.

Winner: Google Kubernetes Engine

Community Winner: Google Kubernetes Engine

South West / Bottom Left

Amazon SNS & Amazon SQS vs Google Pub/Sub

The first unfair match up as SQS and SNS take away from each other. Feature-wise, they line up favourably against Pub/Sub but the breadth of the Pub/Sub feature set wins out…despite naming one of the communications models, “Streaming Pull.”

Winner: Google Pub/Sub

Community Winner: Amazon SQS & SNS

AWS Step Functions vs Azure Logic Apps

Azure Logic Apps brings a far more polished user experience to this match-up. In fairness, AWS Step Functions is a lot more versatile. While the goal is the same, these services target different users…for now. Azure Logic Apps wins this round based on the simplicity of the experience and the speed with which you can get up and running…for now.

Winner: Azure Logic Apps

Community Winner: AWS Step Functions

Azure DevOps vs AWS CloudFormation

Azure DevOps loses this one based purely on the name. DevOps is a philosophy and team strategy, marketing hype need not apply. AWS CloudFormation wins by default.

Winner: AWS CloudFormation

Community Winner: AWS CloudFormation

Google Firebase vs AWS AppSync

Similar, AWS AppSync is the worst named service in a sea of horrible names (just ask Corey Quinn…or don’t, we don’t need to get him started again 😉). The assumption when people hear AWS AppSync is that it’s a data synchronization tool. It’s really a GraphQL service (which is amazing) who’s name does it a disservice. Google Firebase doesn’t even need to show up. Another win by default.

Winner: Google Firebase

Community Winner: AWS AppSync

North East / Top Right

Amazon DynamoDB vs MongoDB Atlas

Kudos for MongoDB Atlas for making the tournament. The seeding round was tough but they made it in. That’s an accomplishment for any service not run by Google, AWS, or Microsoft. Be happy with that.

Winner: Amazon DynamoDB

Community Winner: Amazon DynamoDB

Azure Cosmos DB vs Amazon Aurora

Azure CosmosDB promises to be any type of data store to anyone. What’s more impressive is that they accomplish about 40% of that goal. That’s impressive. Amazon Aurora is great and what I’d choose to build a system on but the promise of Cosmos DB is enough of an edge here. Let’s consider that a buzzer beater for the win.

Winner: Azure Cosmos DB

Community Winner: Amazon Aurora

Google BigTable vs Amazon Athena

Amazon Athena can work wonders but you’ve got to put in a ton of work to get there. Google’s BigTable on the other hand is extremely straightforward but requires spinning up instances. Amazon Athena’s serverless approach slips in behind BigTable’s legacy defence for the win.

Winner: Amazon Athena

Community Winner: Amazon Athena

Google BigQuery vs Snowflake Cloud Data Platform

Another non-big three entry! Congrats to Snowflake for making it in but is that the service or the company or something else entirely? Marketing resiliency under the name of one of the most fragile doesn’t do the underlying platform justice. Google BigQuery doesn’t mess around. Not having to worry about the underlying infrastructure for ridiculously massive queries that execute with lightning speed? Winner.

Winner: Google BigQuery

Community Winner: Google BigQuery

South East / Bottom Right

Amazon S3 vs Azure Blog Storage

Azure Blob Storage always reminds me of the the X-Men villain. It’s a service around giant chunks of data. Amazon S3 can be too. But it’s the original and despite over a decade of service, remains a lean, mean storage machine.

Winner: Amazon S3

Community Winner: Amazon S3

Amazon EFS vs Amazon EBS

Boring vs. more boring. This is a match up for the days before the basketball shot clock. Amazon EFS eeks out the win because it feel more elastic…and was the higher seed going in.

Winner: Amazon EFS

Community Winner: Amazon EBS

Google AutoML vs Amazon SageMaker

Google AutoML gives up a 10 point lead right off the opening tip because the name implies no work needed. Obviously not true but why set that expectation? Amazon SageMaker got a massive boost at AWS re:Invent 2019 and runs away with this one in a blowout.

Winner: Amazon SageMaker

Community Winner: Amazon Sagemaker

Azure Databricks vs Google Cloud Natural Language

The wild card match up. Azure Databricks is a managed Apache Spark service while Google Cloud Natural Language is a natural language processing (NLP) service that deconstructs text. Streamlining a sophisticated computer science problem down to a simple API call? Google Cloud Natural Language runs away with this one.

Winner: Google Cloud Natural Language

Community Winner: Azure Databricks

How Did I Do?

For round #1, I chose 9 winners and lost 7 of the match-ups. That puts me at 56%…yikes. Time to ask the community why they went a different way! Let’s hope that I have a higher winning percentage in the second round.

Of note, I haven’t changed my round #2 predictions based on the failures of round #1. That means my winning percentage is probably going to drop 😞.

Round #2

North West / Top Left

Amazon EC2 vs AWS Lambda

The majority of compute against the up and comer. In reality these services are very complimentary but only one can win in Cloud Madness. This one goes to double overtime but EC2 can beat the event driven approach of Lambda

Winner: AWS Lambda

Community Winner: AWS Lambda

Google Cloud Run vs Google Kubernetes Engine

Kubernetes has all of the hype. It’s synonymous with containers and everyone is expected to need it. Guess what? You probably don’t need it and while you were taken in by the hype, Cloud Run and its pragmatic approach to problem solving was getting buckets.

Winner: Google Cloud Run

Community Winner: Amazon ECS (in an Amazon ECS vs Google Kubernetes Engine match-up)

South West / Bottom Left

Google Pub/Sub vs Azure Logic Apps

Azure Logics Apps looks slick and it solves real business problems. Still, it’s not enough to overcome the utility of Pub/Sub. You can Pub and Sub! That’s hard to beat 😉. Jokes aside, Pub/Sub makes it easy to connect your distributed system and that why they win this round.

Winner: Google Pub/Sub

Community Winner: Amazon SNS & SQS (in an Amazon SNS & SQS vs AWS Step Functions match-up)

AWS CloudFormation vs Google Firebase

After walking through the first round, CloudFormation faces a real test in the second round. Firebase is an amazing service that covers so much ground. However, CloudFormation’s focus and utility—despite its rough edges—wins out in this round.

Winner: AWS CloudFormation

Community Winner: AWS CloudFormation (in an AWS CloudFormation vs AWS AppSync match-up)

North East / Top Right

Amazon DynamoDB vs Azure Cosmos DB

Reality hits hard for Cosmos DB in the second round as they run head first into the data juggernaut that is DynamoDB. DynamoDB was the first, truly new data store at scale and it has only sped up from there. DynamoDB has continued to break records and runs Cosmos DB right out of the gym.

Winner: Amazon DynamoDB

Community Winner: Amazon DynamoDB (in an Amazon DynamoDB vs Amazon Aurora match-up)

Amazon Athena vs Google BigQuery

A true head-to-head match up. These services play the same style and this game goes down to the wire. Amazon Athena isn’t as slick but support from Amazon S3 is what puts them over the top here. There’s a huge amount of value locked in S3 buckets and Athena opens that up. It’s only knock? It’s not used enough!

Winner: Amazon Athena

Community Winner: Google BigQuery

South East / Bottom Right

Amazon S3 vs Amazon EFS

EFS made the second round basically on a coin toss. Reality now hits home, hard. In a record setting rout, Amazon S3 walks into the second round playing only the equipment manager, popcorn seller, and an empty chair.

Winner: Amazon S3

Community Winner: Amazon S3 (in an Amazon S3 vs Amazon EBS match-up)

Amazon SageMaker vs Google Cloud Natural Language

A generic machine learning tool belt against a very specific solution. It’s a team of swing players against the top prospect in the upcoming draft. Despite a record setting individual performance, this is a team game and Amazon SageMaker wins this hard fought match.

Winner: Amazon SageMaker

Community Winner: Amazon SageMaker (in am Amazon SageMaker vs Azure Databricks match-up)

How Did I Do?

With only a 56% win rate for round #1, a good portion of my round two match-ups simply didn’t happen. That said, I somehow did better this round! I ended up with a 62.5% win percentage by picking five of the eight winners.

For round #3, I haven’t changed my predictions at all. We’ll see how my predictions hold up as the madness continues.

Round #3

North West / Top Left

AWS Lambda vs Google Cloud Run

Two different approaches come head to head in this round. Google Cloud Run aims to provide serverless benefits to lightly modified containers. AWS Lambda is the core of serverless bringing a modern approach to compute. While the pragmatic Cloud Run approach is fantastic, AWS Lambda wins this one with a buzzer beater.

Winner: AWS Lambda

Community Winner: AWS Lambda (in an AWS Lambda vs Amazon ECS match-up)

South West / Bottom Left

Google Pub/Sub vs AWS CloudFormation

Later rounds of the tournament are always interesting as these matches are where very different approaches come head to head. Pub/Sub addresses a core problem with distributed systems while CloudFormation is critical to repeatable ops. This match up goes down to the wire but CloudFormation’s delayed support for new services means Pub/Sub ekes out a win

Winner: Google Pub/Sub

Community Winner: AWS CloudFormation (in an AWS CloudFormation vs Amazon SNS & SQS match-up)

North East / Top Right

Amazon DynamoDB vs Amazon Athena

The excitement of the newcomer that flew under the radar in Athena against the proven juggernaut of DynamoDB. Athena’s potential to unlock data in S3 is unbounded but is it enough to take down a proven winner in DynamoDB? A service that is running massive workloads across the world 24/7? No. Cloud Madness is where dreams come true but this match up closes the book on Athena’s dream.

Winner: Amazon DynamoDB

Community Winner: Amazon DynamoDB (in an Amazon DynamoDB vs Google BigQuery match-up)

South East / Bottom Right

Amazon S3 vs Amazon SageMaker

Machine learning has broken through the hype cycle and is delivering real results. With a massive upgrade at AWS re:Invent 2019, SageMaker has come to win. But you can’t use machine learning if you don’t have the data and Amazon S3 is quite possibly the single biggest repository of data on the planet. That weight can’t be ignored.

Winner: Amazon S3

Community Winner: Amazon S3

How Did I Do?

100%. Straight up nailed it.

Semifinals

AWS Lambda vs Google Pub/Sub

After a bit of a scare in the last round, AWS Lambda has recovered and is set to tackle any challenge. Google Pub/Sub continues to drive a diverse set of architecture including serverless ones. But Lambda can’t be denied. It came out of the gate as a never seen before service and has started a revolution in cloud native designs. That momentum carries it through to the final.

Winner: AWS Lambda

Community Winner: AWS Lambda (in an AWS Lambda vs AWS CloudFormation match-up)

Amazon DynamoDB vs Amazon S3

DynamoDB’s revolutionary approach against S3’s simplicity means these services are trading bucket for bucket. This match goes to double overtime. While simple, S3 legacy permissions features are confusing and a knock against them. DynamoDB isn’t without its challenges, as single table design is tough for teams to wrap their head around. Down to the end of the bench, after an exhausting grudge match, S3’s “I put data in bucket” simple approach won’t be denied.

Winner: Amazon S3

Community Winner: Amazon S3

How Did I Do?

Another perfect round. It’s very weird that my winning percentage has gotten better each round but I’m not complaining! Let’s see how the finals play out.

The Finals

AWS Lambda vs Amazon S3

Can we call it a draw and walk away? After all, Amazon S3 triggers invoking AWS Lambda functions is a powerful combination. Sadly, that’s not how Cloud Madness works.

Amazon S3 kicked off the cloud revolution and has stayed focused ever since. It’s only knock, that legacy permissions structure that leads to some costly data turnovers for some users.

AWS Lambda came out of nowhere five years ago and shocked the cloud world. The serverless community simply wouldn’t exist without this service. It’s trick play, glue—not that glue—for your cloud deployment. No need to be a cutting edge cloud user to see the value.

In a match that’s all offence, AWS Lambda wins on a corner three to walk away cloud champion! It’s impact on the broader community pulling through in the end.

Community Winner: Amazon S3

How Did I Do?

After getting thumped in the first round, I made a strong come back and performed well all the way up to the finals. In a surprise to most, Amazon S3 comes out on top. Fair enough, S3 is the original serverless service and has the best track record on the planet for reliability and longevity.

Here’s the official summary from A Cloud Guru and Forrest’s post about the champion.

Congrats Amazon S3!

Your Turn

We’ll see how accurate my bracket it as Cloud Madness plays out over the month. Here’s my completed bracket:

@marknca's completed Cloud Madness bracket showing the winners of each successive round

Win my guesses be spot on? Completely wrong? Honestly, I’m just excited to see how this plays out over the month. The community will be voting on each match up over on the A Cloud Guru social channels. Be sure to follow Forrest and the A Cloud Guru team to participate!


If you didn't already know, I've been involved with the team at A Cloud Guru for quite a while now. They employ a number of other AWS Heroes and the team does fantastic work. Of course, I'm little biased in that opinion because I've published two courses on the platform!