What does "event source length" mean?

Question

What does "event source length" mean?

Gallatin 21V 301

Hi Team，

https://docs.azure.cn/en-us/azure-functions/functions-target-based-scaling?tabs=v5%2Ccsharp

User's image

According to the document, the expansion of the number of function instances in the EP pricing layer is carried out according to the above formula.

The trigger type of the function is eventhub trigger. Then, does the event source length is the number of incoming messages in the eventhub?According to my calculation, it seems that it is not "incoming messages" in eventhub.

Anonymous

2025-06-25T15:21:58.5866667+00:00
Hi Gallatin 21V,

Thanks for your question.

You're correct that the function instance scaling formula in the Elastic Premium (EP) plan is based on the event source length. However, for Event Hub triggers, the term “event source length” does not directly refer to the total number of incoming messages in the Event Hub.

Instead, it refers to the number of unprocessed events that are available in the Event Hub partition(s) that the function is monitoring. This value is influenced by:

The current lag (messages not yet checkpointed),

The number of partitions, and

The function’s processing throughput.

So if your function is fast enough to process incoming messages, the "event source length" may remain low even if the Event Hub is receiving a high volume of messages.

If you're seeing differences between the actual number of incoming messages and what triggers scaling, it's likely because the unprocessed event backlog (i.e., the source length) is not growing significantly — meaning the function is keeping up well with the incoming load. To help you better understand, kindly refer to the documentation below.

Event-driven scaling in Azure Functions

I hope this is helpful! Feel free to reach out if you have any more questions or need further assistance.
Anonymous

2025-06-26T14:34:03.9466667+00:00

Hello Gallatin 21V,

I hope you had a chance to review the response above. If there are any updates or if you need further assistance, please feel free to update here.
Gallatin 21V 301 Reputation points

2025-06-27T08:07:21.4633333+00:00

Hi @Anonymous

Thank you for your reply. However, how can we calculate the number of unprocessed messages that are accumulated in the EventHub? I understand it should be "incoming messages" minus "outgoing messages".Am I correct in my understanding?
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-10T10:54:35.2933333+00:00

Hello Gallatin 21V •,
You're correct — Unprocessed messages can be estimated as: Unprocessed = Incoming Messages - Outgoing Messages

However, for a more accurate view, it's better to use Event Hub Consumer Group Lag via Azure Monitor. This shows the actual number of events pending for each consumer group.
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-11T05:09:51.16+00:00

Hello Gallatin 21V •,
Just checking in to see if you've had a chance to review my previous response. Let me know if you have any additional questions.
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-14T05:43:34.5233333+00:00

Hello Gallatin 21V •,
I wanted to follow up to see if you’ve had an opportunity to look over my earlier response. Please let me know if you have any further questions.
Gallatin 21V 301 Reputation points

2025-11-07T09:48:47.1566667+00:00

Hi @Krishna Chowdary Paricharla Unprocessed = Incoming Messages - Outgoing Messages. This seems incorrect.

For example, if an event hub receives 10,000 messages in a short period of time and the consumption is very slow, then the incoming messages metric will quickly become 0, but the outgoing messages continue to be ongoing, resulting in a negative value.

You mentioned that by looking at Azure Monitor, the Mooncake event hub portal cannot configure insights. So, what exactly are you referring to?

Your answer

Anonymous

2025-06-25T15:21:58.5866667+00:00

Hi Gallatin 21V,

Thanks for your question.

You're correct that the function instance scaling formula in the Elastic Premium (EP) plan is based on the event source length. However, for Event Hub triggers, the term “event source length” does not directly refer to the total number of incoming messages in the Event Hub.

Instead, it refers to the number of unprocessed events that are available in the Event Hub partition(s) that the function is monitoring. This value is influenced by:

The current lag (messages not yet checkpointed),

The number of partitions, and

The function’s processing throughput.

So if your function is fast enough to process incoming messages, the "event source length" may remain low even if the Event Hub is receiving a high volume of messages.

If you're seeing differences between the actual number of incoming messages and what triggers scaling, it's likely because the unprocessed event backlog (i.e., the source length) is not growing significantly — meaning the function is keeping up well with the incoming load. To help you better understand, kindly refer to the documentation below.

Event-driven scaling in Azure Functions

I hope this is helpful! Feel free to reach out if you have any more questions or need further assistance.
Anonymous

2025-06-26T14:34:03.9466667+00:00

Hello Gallatin 21V,

I hope you had a chance to review the response above. If there are any updates or if you need further assistance, please feel free to update here.
Gallatin 21V 301 Reputation points

2025-06-27T08:07:21.4633333+00:00

Hi @Anonymous

Thank you for your reply. However, how can we calculate the number of unprocessed messages that are accumulated in the EventHub? I understand it should be "incoming messages" minus "outgoing messages".Am I correct in my understanding?
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-10T10:54:35.2933333+00:00

Hello Gallatin 21V •,
You're correct — Unprocessed messages can be estimated as: Unprocessed = Incoming Messages - Outgoing Messages

However, for a more accurate view, it's better to use Event Hub Consumer Group Lag via Azure Monitor. This shows the actual number of events pending for each consumer group.
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-11T05:09:51.16+00:00

Hello Gallatin 21V •,
Just checking in to see if you've had a chance to review my previous response. Let me know if you have any additional questions.
Krishna Chowdary Paricharla 2,915 Reputation points Microsoft External Staff Moderator

2025-07-14T05:43:34.5233333+00:00

Hello Gallatin 21V •,
I wanted to follow up to see if you’ve had an opportunity to look over my earlier response. Please let me know if you have any further questions.
Gallatin 21V 301 Reputation points

2025-11-07T09:48:47.1566667+00:00

Hi @Krishna Chowdary Paricharla Unprocessed = Incoming Messages - Outgoing Messages. This seems incorrect.

For example, if an event hub receives 10,000 messages in a short period of time and the consumption is very slow, then the incoming messages metric will quickly become 0, but the outgoing messages continue to be ongoing, resulting in a negative value.

You mentioned that by looking at Azure Monitor, the Mooncake event hub portal cannot configure insights. So, what exactly are you referring to?