For a deeper look into our DataScope Select SOAP API, look into:

Overview |  Quickstart |  Documentation |  Downloads |  Tutorials

question

Upvotes
Accepted
462 9 7 14

DSS REST API - 'On Demand' vs. 'Schedule' - any performance differences?

Hi,

When using DSS REST API in terms of performance (speed of extraction) is there any known/expected difference between using an 'On Demand' (data returned in response) and 'Schedule' ('physical')? In summary the question is, does one method return the data more efficiently than the other?

Many thanks,

Gareth

dss-rest-apidatascope-selectdss
icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

Upvote
Accepted
13.7k 26 8 12

@gareth.teage

I do not see any reason why a scheduled or On Demand request would have different performance. My reasoning is that the main time consumer is the data extraction process itself, not the way it was triggered.

It is true that scheduled requests are triggered directly by the server, whereas On Demand requests are queued, and if the queue is very busy, there might be a delay before the extraction starts. But how meaningful is that delay, statistically, how does it compare to the extraction time ? Note also that if you have to create an instrument list, template and schedule, those 3 steps also take time … compared to a single On Demand request.

My choice between scheduled and On Demand would be based on the workflow and use case, not on a performance consideration, but that’s just an opinion.

As Zoya mentioned, extraction times vary, based on a whole set of factors, like server load.

That said, if you observe occasional cases where the extraction time is significantly higher compared to usual values, there are several things you can do to understand why it drops:

  • Check if there is a service alert at the same time. The best way to find service outage information is to subscribe to alerts and notifications for the product(s) of interest to you.
    You can also view them in the ALERT page in Eikon, or in My Account.
  • Check the extraction notes.
  • Add a unique Client-Session-Id to each request. This must be programmed in advance, as part of the standard diagnostics mechanisms of your application; it cannot be done after an issue occurred. For more information on this ID (and the following), see this Diagnostics page.
  • Log the Request-Execution-Correlation-Id from the response header of the last result's message.
  • If you want us to investigate why a specific call had significantly lower performance than usual, contact customer support and give them the Client-Session-Id and the Request-Execution-Correlation-Id for the specific call. This will help us for our research on the server side.
icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

Upvote
23k 22 9 14

No. This should not be happening. So please use the way to request that you prefer.

What you should expect to see, is that processing time may vary. The exact same request is not getting the exact same processing time on different days, and at different times of the day.

icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

Click below to post an Idea Post Idea