How Canva Saved Millions in AWS S3 Costs

Summary

Canva, an online design platform with over 100 million monthly users, has accumulated over 230 petabytes of data on Amazon S3. To optimize costs, Canva analyzed their data access patterns and discovered that transitioning 90% of their data to S3 Glacier Instant Retrieval could save $1.8 million monthly. However, they had to consider a one-time transition fee and the minimum billable object size. After careful analysis, Canva prioritized migrating larger objects, achieving significant cost savings and reducing monthly expenses by approximately $300,000.

Takeaways

• 🚀 Canva has experienced explosive growth, now serving over 100 million active monthly users who have created more than 15 billion designs.
• 📊 This growth has led to a massive data accumulation of over 230 Petabytes in Canva's Amazon S3 storage.
• 💵 Canva's potential monthly storage costs on Amazon S3 are substantial, with the first 50 TB costing $1.2k and the next 450 TB costing nearly $5 million.
• 🔗 Amazon S3 is a cloud storage solution that allows users to store data as objects in buckets with unique URLs for access.
• 🗂️ S3 offers features like versioning, access permissions, and automatic scaling to handle data from gigabytes to petabytes.
• 📈 Canva analyzed their data usage and found that user-generated content was frequently accessed within the first 15 days but less so afterward.
• 💼 Canva considered transitioning to S3 Glacier Instant Retrieval for its infrequently accessed data, which could lead to significant cost savings.
• 💸 The potential savings for migrating 207 PB of data from S3 Standard-IA to S3 Glacier Instant Retrieval could be $1,842,300 per month.
• ⚖️ Transitioning to S3 Glacier Instant Retrieval involves a one-time fee of $0.02 per 1000 objects, which could cost over $6 million for Canva's 300 billion objects.
• 📉 Canva decided to prioritize transitioning objects with an average size of 400KB or larger to S3 Glacier Instant Retrieval for a quicker return on investment.
• 🔧 Canva implemented lifecycle policies to automate the migration of objects to S3 Glacier Instant Retrieval, successfully migrating nearly 80 billion objects in two days.

Q & A

What is Canva and how has its user base grown since 2013?

Canva is an online design tool that has experienced astronomical growth since 2013, now serving over 100 million active monthly users who have created more than 15 billion designs.

What is the scale of data accumulation on Canva's Amazon S3 storage?

Canva has accumulated over 230 Petabytes of data across its Amazon S3 storage, illustrating the massive scale of its operations.

What is Amazon S3 and how does it benefit users?

Amazon S3 is a cloud storage solution that allows users to upload data as objects into containers called buckets. It offers benefits such as versioning, access permissions, policies, and automatic scaling to handle any amount of data, charging users for only what they use.

What are the different storage classes offered by Amazon S3 and their purposes?

Amazon S3 offers various storage classes including S3 Standard for frequently accessed data, S3 Standard-IA for less frequent access data, and S3 Glacier Instant Retrieval for infrequent access with fast retrieval when needed.

How did Canva use S3 Storage Class Analysis to optimize its storage costs?

Canva used S3 Storage Class Analysis to gain insights into their data usage and found that much of their data fit the profile for S3 Glacier Instant Retrieval, which offers lower storage costs with fast retrieval times.

What are the potential savings for Canva if they migrate 207 PB of data to S3 Glacier Instant Retrieval?

The potential savings for Canva by migrating 207 PB of data to S3 Glacier Instant Retrieval could be $1,842,300 per month or $22,107,600 per year, compared to the cost of S3 Standard-IA.

What is the one-time fee associated with transitioning objects between storage classes on Amazon S3?

There is a significant one-time fee for transitioning objects between storage classes on Amazon S3, which is $0.02 per 1000 objects.

What considerations did Canva take into account before transitioning to S3 Glacier Instant Retrieval?

Canva considered the minimum billable object size of 128KB for S3 Glacier Instant Retrieval, the one-time transition cost, and the time it would take to break even when moving from S3 Standard or S3 Standard-IA to S3 Glacier Instant Retrieval.

What strategy did Canva implement to prioritize the migration of objects to S3 Glacier Instant Retrieval?

Canva decided to prioritize transitioning objects with an average object size of 400KB or larger to S3 Glacier Instant Retrieval to ensure a positive return on investment within 6 months.

How did Canva implement their storage optimization strategy and what was the result?

Canva applied lifecycle policies to the targeted S3 buckets, which automatically migrated objects to S3 Glacier Instant Retrieval based on predefined rules. This process allowed them to successfully migrate nearly 80 billion objects in approximately two days, leading to significant cost savings.

What was the impact of Canva's storage optimization on their S3 storage expenses?

After the migration, Canva reduced their S3 storage expenses by roughly $300,000 per month or $3.6 million per year as they progressively migrated objects over to S3 Glacier Instant Retrieval.