Testing the scalability of Microsoft Dynamics GP 10.0…


With every release, and sometimes multiple times within a release, our test team takes a crack at proving out just how scalable Microsoft Dynamics GP really is when put under some demanding scenarios…demanding, as in 1,000 concurrent users demanding!  Well they just completed another round of this testing on our 10.0 SP2 release, along with both Microsoft SQL Server 2008 and Windows Server 2008. The results are impressive to say the least!  A whopping 3.28 MILLION business transactions and 10.66 MILLION business transaction line items were processed over the period of an 8 hour day!

Partners can download the full report here, but I’ll give everyone some details below on just how well Dynamics GP can perform when put to the test…

The first thing you have to realize is that we run this test multiple times to validate the results, and we disperse the 1,000 concurrent users throughout the system to mimic a real-world type of scenario.  Here is the breakout of how the users, transactions, and processes are split out:

Transaction Type Transaction Line Count Number of Clients/Users
General Ledger Transaction Entry 10 115
Payables Voucher Entry 1 100
Receivables Cash Entry 1 105
Purchase Order Entry 5 50
Purchase Order Received 5 25
Purchase Order Posted 5 25
Sales Order Entry 5 400
Sales Order Transfer 5 75
Sales Order Posting 5 75
Receivables Cash Posting N/A 20
Payables Voucher Posting N/A 5
Payables Historical Aged Trial Balance N/A 1
Receivables Historical Aged Trial Balance N/A 1
Payables Check Processing N/A 1
Receivables Month End N/A 1
US Payroll Check Processing N/A 1
Total Constant Concurrent Users   1,000

Now, with that level of activity going on in the system, here is the throughput of each area when tested over the period of a typical 8 hour workday:

Transaction Type Transactions per hour Transaction lines per hour
General Ledger Transaction Entry 13,978 139,780
Payables Voucher Entry 35,589 35,589
Payables Voucher Posting 74,697 74,697
Receivables Cash Entry 68,617 68,617
Receivables Cash Posting 27,389 27,389
Purchase Order Entry 4,473 23,715
Purchase Order Received 6,170 30,850
Purchase Order Posted 24,565 122,825
Sales Order Entry 49,459 247,295
Sales Order Transfer 49,441 247,205
Sales Order Posting 51,049 255,245
US Payroll Check Processing 5,000 60,000

Do the math yourself.  This test proved Microsoft Dynamics GP can process up to 3.28 MILLION business transactions and 10.66 MILLION business transaction line items over an 8 hour day!

Pretty incredible if you ask me. And a very nice improvement over the previous test which leveraged GP 10.0 RTM with SQL Server 2005 and Windows Server 2003 (the test that delivered 2.6M transactions with 8.3M transaction lines).  This new test actually used the SAME hardware as the previous 1,000 concurrent user test, with the same test methodology and data.  Which means in an apples-to-apples comparison, SQL Server 2008 and Windows Server 2008 showed a whopping 26% improvement in transaction throughput (+28% in transaction line throughput)!  That is some serious performance!

AndyW

Comments (9)

  1. rdrag says:

    I appreciate these stats.  GP has come a long way in improving performance, due to both better hardware (hooray for SQL Server) and improvements to the Dexterity base code.

    But one thing I lament is that this bench test is focused on GP users that only do a few line items per sales invoice.  Do you folks ever through transactions with a few thousand line items into the mix?  We have a client who, believe it or not, regularly enters orders with up to two thousand line items.  That many line items, in combination with applying a trade discount, really affects system performance.  I suggest running some tests like that, if there are opportunties for improvement.

  2. Steve Gray says:

    Do you have any figures that pertain to eConnect?

  3. gibson.claire@gmail.com says:

    Thanks so much for this.

    I also agree with the comment that bench marking on multiple line items needs to be done.

    This benchmark is also only aimed at transactions created directly within GP. Are there any plans to test eConnect integrations with this amount of transactions? I have a client running BizTalk and eConnect and we are having some interesting challenges around integrating and then posting large amounts of transactions over a half hour period. It would be nice to go back to them with some kind of stat because they are now investigating other products.

  4. Great ideas on the transaction line items and eConnect,

    The scalability test’s goal is to test a typical customer scenario, and most do not have thousands of line items.  Therefore, I doubt we would change the test, but I will forward the comment and see if we can broaden it for future tests.

    We see more and more customers integrating with eConnect, Web Services, and Integration Manager (as well as other 3rd party integration products.)  A performance study has not been done on these tools, but I think it would be wise to do so.  We’ll see what we can do.

    Pam

  5. Martin Olsen says:

    With eConnect and similar tools the speed of integration is great. The pain point then becomes the time taken to post these large batches (and the fact that this can only be done with a GP client up and running, so much more difficult to do overnight). I would be interested in the tests performed on posting batches with thousands of transactions. With GP often being the recipient database of transactions from multiple front end systems the posting of large batches is becoming much more important.  

  6. Ryan Weigel says:

    As most of you are probably aware we will be releasing the next version of GP in 2010. Around the release of that we will again be doing another large benchmark/scalability test of some sort. I will be involved in the planning of that. So I will consider the requests mentioned above when we get into the planning of this. I can’t make any for sure promises but we will definitely look into these areas for consideration. Machine availability and test resource availability play a big role into what we can all test. I appreciate all the feedback on this as it is always nice to know what people would like to see from a performance perspective.

  7. gibson.claire@gmail.com says:

    Hi Steve,

    I am in the process of collecting some accurate stats and will revert soonest.

  8. Joanna Shaffer says:

    Have you considered including Project Accounting in any of your performance & scalability testing?

  9. Thanks for the feedback Joanna!  I’ll pass this along to our test team for consideration in their next round of Dynamics GP Scalability testing.

    AndyW

Skip to main content