Best Practices

To optimise the performance of centron S3 Object Storage, based on your use case and architecture, you should follow these recommendations:

Use of a Content Delivery Network (CDN):

When: Particularly useful when handling mainly GET requests from the internet, especially where there are frequent requests for small files.

Advantages: Content distribution with low latency and a higher data transfer rate. Reduction in latency and fewer GET requests to centron and S3 Object Storage.

Implementation: Use of an upstream CDN with minimal configuration required.

Best practice for naming objects:

When: Recommended if you expect to use the ListObject API call and store a significant number of objects in your bucket.

Advantages: Faster performance of the ListObject API when object keys have different prefixes.

Implementation: Use a prefix for each object key name in the bucket, e.g. random characters or the current date.

Avoiding small files and using multi-part uploads for large files:

When: When working with files smaller than 1 MB or uploading files larger than 500 MB.

Advantages: A reduction in the total number of enquiries to your Space.

Implementation: Use multi-part uploads for files larger than 500 MB and combine smaller files into a single larger file

Choosing the right data centre for your resources:

When: Depending on the origin of the connections to your buckets.

Advantages: Minimal latency and a stable connection via the centron backbone.

Implementation: Select regions for your resources during set-up or migrate existing infrastructure.

Correct handling of 50x errors:

When: Every time a file is uploaded to centron S3 Object Storage.

Advantages: Faster uploads and less manual intervention.

Implementation: Use of S3-compatible clients or libraries with retry logic.

Optimising your enquiry rate:

When: If there are more than 150 requests per second or if 503 ‘Slow Down’ responses occur frequently.

Advantages: Avoids unexpected throttling.

Action: Combine small files or open a support ticket if there is a high volume of enquiries.

Using local or block storage instead of object storage:

When: For dynamic or structured data that requires traditional file system access.

Advantages: Improved performance for specific use cases thanks to low-latency I/O that is closely integrated with the hardware.

Implementation: Start with volumes or use database solutions such as Redis or Cassandra.