Throughput

Are EdgeKV operations performed from an EdgeWorkers function throttled?

Throttling is enforced when requests need to go to the persistent store. As such, all write and delete requests are subject to the throughput limits listed in the EdgeKV product limits.

For read requests, the throughput limit is only enforced:

  • On requests for items that do not have a local copy on the edge server executing the EdgeWorkers function making the request.
  • Or when the requested items were invalided due to a recent write or delete.

How does the read/write ratio impact my throughput limit?

When you update an item by performing a write or delete operation, subsequent read requests for that item have to go to the persistent store. There is also an increased chance that subsequent read requests of other items have to go to the persistent store. When a request goes to the persistent store the throughput limit is applied.

Minimizing the update frequency will reduce the chances of hitting the throughput limit.

How can I drive more traffic while staying within the EdgeKV throughput limits?

To reduce the chance of hitting the throughput limit:

  • Maintain a read/write ratio of 98/2 read/write or better.
  • Avoid long tail data requests (data that is infrequently accessed).

Are the EdgeKV request limits enforced globally?

Yes, the limits are enforced globally.

How can I perform small operations?

You can use the EdgeKV CLI or API to perform small data operations. If you use the EdgeKV CLI or API to perform a bulk-data operation the request might be throttled due to the significant load.

How can I perform bulk data operations?

To perform a bulk data operation, you can use an EdgeWorkers function to send the requests. EdgeWorkers support high volume operations and high limits. You can create an EdgeWorker and call it several times to perform bulk write operations. You can also call an EdgeWorker several times to request all the values in a database quickly and without being throttled. To perform these requests you need the itemID parameter for each value in the database.

What if I can’t perform my bulk operation using EdgeWorkers?

If you are unable to use EdgeWorkers to perform bulk operations you can contact your Akamai account team for assistance. Please prepare the following information so an Akamai representative can perform a bulk-data operation on your behalf.

  • The database and namespace for the data operation.
  • A secure way to transfer the data to your account representative. For example, a Webex contact or an email address.