When Digital Ocean announced the availability of a managed Postgres database service, I wanted to answer this question:

Would my app’s performance improve if I switched my Postgres instance from AWS to Digital Ocean?

I'm working on Medtally.com, a data-driven medical community. Because nearly every page on Medtally requires fresh data, the speed of the database is crucial to providing a fast experience for users.

Example page on Medtally. Every page request pulls data from a Postgres database.

Thanks to the generosity of Digital Ocean's Hatch program, I was able to get some free credits to run a test.

Test setup:

AWS (original): Postgres 11 (2 vCPUs, 4GB RAM, 40GB storage, AWS us-west-1)

Digital Ocean: Postgres 11 (2 vCPUs, 4GB RAM, 38GB storage, SFO2)

App hosting: Zeit Now v1 (SFO, AWS us-west-1)

App stack: Express.js, React.js, Next.js (read my tips on building with Next.js)

How performance was measured:

To measure performance as close to the real-world as possible, I tracked the response time of a common query required to fetch data for our popular health condition pages (example).

I used Node.js' perf_hooks API and logged the results.

Here's the example code I used:

const { performance } = require('perf_hooks');
const t0 = performance.now(); //start timer
const data = await fetchDataFromDB(id);
const t1 = performance.now(); //end timer
const responseTime = t1 - t0 //responsetime in ms

Results after ~2,000+ samples:

Average response time:

• AWS: 145ms
• Digital Ocean: 233ms

Median response time:

• AWS: 63ms
• Digital Ocean: 164ms

Digital Ocean's response time was 60% longer on average and 160% by median.

The app felt noticeably slower while I clicked around. The Medtally app can load pages in as little as 150ms. Adding 100ms to that time would mean 66% slower - a significant increase.

I was surprised. I expected the performance to be much closer since both databases had the same number of CPUs, RAM size, and disk size.

Some possible reasons why AWS was faster:

• The AWS Postgres instance had an advantage in latency because it was hosted in the same physical location as my app (AWS us-west-1 which is San Francisco). However, Digital Ocean's database was also hosted in San Francisco and so the difference should have been negligible and should not have been a 100ms.

Update: The difference is only 2ms after I ran a few pinging tests between the AWS us-west-1 and Digital Ocean SFO2. We can rule this reason out.

Note: I could have eliminated latency as a variable in the test by basing the conclusion on query execution times in pg_stat_statements instead. I did not have enough time to do this but I may revisit in the future.

• AWS Postgres simply uses faster hardware (CPU, RAM, SSD).

• Digital Ocean's managed Postgres service is new and the company needs more time to optimize its performance. AWS RDS has been in service for 9 years compared to less than a year for Digital Ocean.

Back to AWS RDS

I will be sticking with AWS RDS until Digital Ocean improves its performance.

Besides the performance issues, I really did enjoy using it. The UI/UX is far superior to AWS and so is the documentation. I have high hopes for it in the future and we could use some strong competition in managed database hosting.

With that said, Medtally.com is back to using AWS RDS.

---

Interesting in connecting with me? You can find me on Twitter.