We currently have a .NET application that retrieves/updates data in Salesforce using the API. We are encountering long processing time where the session can time out as well as a bad user experience. We are using filters in the soql queries, but in many cases the fields are not indexed, so the retrieval is slow. I know we can request fields to be indexed by Salesforce to help alleviate the performance drag.

My question is would an APEX web service provide better performance? If we created an APEX class with web service methods that the .NET application could call, should the performance be better theoretically? If so, what options are there to secure that connection? I would want to make sure the connection between the .NET application and Salesforce was secure.

Thanks for any help. Regards.

2 Answers 2


To address the last question first, your connection to salesforce.com is always secure. The SOAP bindings always default to 'https', so you automatically get the best cryptographic tunneling your language supports. For .Net, I'd suppose you'd probably get TLS 1.2 or some such (since TLS 1.3 is still only in draft form). This is true even if you have to proxy or go through firewalls or VPNs. Nothing more than the IP address of the server you're trying to reach is exposed. You get this behavior "for free," without any customization or modification of the code or WSDL.

Now, for the other question. The Apex Code engine and the SOAP/REST API calls all rely on a single query engine that performs automatic optimizations, query execution planning, and so on. However, there's also intrinsic limits within Apex Code that make it particularly unsuited for some scenarios, especially if you may need to return more than 50,000 rows, etc.

If your query is timing out through the SOAP API directly, it's equally likely that your transaction will time out when using the webservice keyword. You will not gain any performance, and you'll lose some fundamental features, such as the ability to page beyond 50,000 rows of data. Additionally, you'll add the execution overhead of loading your class and executing it, which may add a non-trivial amount of time to your overall transaction time, as opposed to using an API call directly, depending on how much data you need to serialize/deserialize, etc.

However, all that being said, there is still a slim chance that moving to an Apex Code Web Service call might improve overall performance if you make some carefully calculated changes. For example, you might get a boost if you use "without sharing" and exclude some filters directly in the SOQL query, and filter them out post-query. This assumes that your newly modified query can return less than 50,000 rows and the time spent filtering in Apex Code doesn't exceed the time the database spends doing the same basic behavior. Such a change may be possible, but the odds are against you.

Personally, I would opt for making a field indexed, either by External ID or by custom index, than try to get around not having an index through code. That's almost always a Bad Idea. Even better, if you find you use certain combinations of filters frequently, consider making a workflow rule that concatenates those values into a single indexed field. Also, you might be able to leverage a SOSL call with additional filters to improve performance. This solution is somewhat limited, as it can't find records until they are indexed, nor can the initial search term be limited to a specific field (only a class of fields, such as phone, email, or all fields).

Your specific use case will determine what the best solution may be, but I would suggest simply trying it out by making a simple mockup and see if there's any average performance gain. This would best be done in a full copy sandbox, of course, to make sure you're getting the same basic behavior to begin with.

  • I just love your answers man. Brilliant explanation. sfdcfox, are you on linkedIn or facebook? May 28, 2015 at 5:26
  • @heisenberg Sure. You can connect with me via any channel you feel like from about.me/brian.m.fear.
    – sfdcfox
    May 28, 2015 at 5:43

The Apex web service would only make a significant performance improvement if:

  • It replaced multiple round trip calls from the .NET application to Salesforce
  • Transferring the compressed payload as JSON rather XML made a significant difference.

For relatively small amounts of data, you won't see much difference in transferring the compressed data in JSON vs SOAP. The most important thing here is that you are using the transport compression on the response.

It would be best to measure the performance of the various parts of your process and then focus on the slowest parts first.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .