![Memory Management in SharePoint 2007 Development Matt Vignau Liquid Hub, Inc www.liquidhub.com [email_address]](https://image.slidesharecdn.com/matthewvignau-memorymanagementinsharepoint2007development-100803223431-phpapp02/85/Matthew-Vignau-Memory-Management-in-SharePoint-2007-Development-1-320.jpg)
Matthew Vignau: Memory Management in SharePoint 2007 Development
- 1. Memory Management in SharePoint 2007 Development Matt Vignau Liquid Hub, Inc www.liquidhub.com [email_address]
- 2. About me Matt Vignau
- 3. Working with SharePoint since SP 2003
- 4. Work for Liquid Hub
- 6. Overview In C# .NET development we have the Garbage Collector
- 8. SPSite and SPWeb not always disposed
- 9. Leaves growing used memory block
- 11. Memory Usage Available blocks of space
- 13. 64 bit vs 32 bit
- 14. Restaurant Analogy Space in memory is reserved, then committed
- 15. Memory is reserved in chunks – like number of diners
- 16. Memory wants to “sit together”
- 17. Memory Leaks Build with each use of the code
- 18. SPSite and SPWeb sites have ~2kB size wrappers
- 19. Object sizes are closer to 1-2 MB
- 20. One site with 15 webs accessed twice has used around 30 MB of system memory; for *one* user
- 21. Can bring down entire farms
- 22. Running in Debug mode
- 23. /3 GB switch – Doesn't work in SharePoint!
- 24. Example 1 public void GetNavigationInfo() { SPWeb OurWeb = http://intranet.litwareinc.com; foreach( SPWeb OurWeb in OurWeb .GetSubWebsForCurrentUser()) { //Our Subsite code here } }
- 25. Memory Danger Signs Does the application pool refresh frequently when under load? Memory reset threshold 800 megs -1.5 gigs Does the system perform poorly under heavy loads?
- 26. Does the entire system crash or users receive “Page not available”?
- 27. Does the system use custom webparts or third party webparts?
- 28. Are you running multiple applications from the same app pool?
- 29. Does the ULS log state “Potentially excessive number of SPRequest threads currently unreleased”?
- 30. Was any compiled code released in debug mode?
- 31. Example 2 public void GetNavigationInfo() { SPWeb OurWeb = http://intranet.litwareinc.com; foreach( SPWeb OurWeb in OurWeb.GetSubWebsForCurrentUser()) { //Our Subsite code here OurWeb.Dispose(); } }
- 32. Example 3 pt 1 try { SPSite OurSiteObject; SPWeb OurObject; //our main code body here }
- 33. Example 3 pt 2 catch { //Our exception handling code }
- 34. Example 3 pt 3 finally { if(OurObject != NULL) { OurObject.Dispose(); } if(OurSiteObject != NULL) { OurSiteObject.Dispose(); } }
- 35. Breakdown Try-catch block allows for exception handling
- 36. Finally block executes the dispose after the code block is complete to avoid problems
- 37. No explicit calls to dispose necessary for SPContext initializations*
- 38. Using statements One of the most efficient means
- 39. Single block usage with automatic disposal
- 40. No additional call out to the dispose method is needed
- 41. Cannot use method outbound transfer of object beyond the using block
- 42. Example 4 using (SPWeb OurWeb = null){
- 43. //Our Code Here;
- 44. }
- 45. Caching Saves on memory
- 47. Is not thread-safe IIS is multi-threaded Best for single-user applications
- 48. Example 5 public void CacheData()
- 49. {
- 50. SPListItemCollection oListItems;
- 51. oListItems = (SPListItemCollection)Cache["ListItemCacheName"];
- 52. if(oListItems == null)
- 53. {
- 54. oListItems = DoQueryToReturnItems();
- 55. Cache.Add("ListItemCacheName", oListItems, ..);
- 56. }
- 57. }
- 58. Example 6 public void CacheData()
- 59. {
- 60. DataTable oDataTable;
- 61. SPListItemCollection oListItems;
- 62. lock(this)
- 63. {
- 64. oDataTable = (DataTable)Cache["ListItemCacheName"];
- 65. if(oDataTable == null)
- 66. {
- 67. oListItems = DoQueryToReturnItems();
- 68. oDataTable = oListItems.GetDataTable();
- 69. Cache.Add("ListItemCacheName", oDataTable, ..);
- 70. }
- 71. }
- 72. Scalability Target scale can impact performance considerations
- 73. Even with proper dispose, memory can grow
- 74. Projected concurrent users are a factor in your design
- 75. Size of farm employed to host
- 76. Scale Questions Pt 1 Is the data static, somewhat static (occasionally changes), or dynamic (often changes)?
- 77. Is the data the same for all users or does it change? (Dependent on user account, department within the company, etc)
- 78. Is the data easily accessible or does it require a long time to return the data?
- 79. Scale Questions Pt 2 Is the data public or does it require a higher level of security?
- 80. What is the size of the data?
- 81. Is the SharePoint site on a single server or on a server farm?
- 82. Links MSDN Best practices : http://msdn2.microsoft.com/en-us/library/aa973248.aspx#
- 83. Memory Pressure in MOSS: http://blogs.technet.com/stefan_gossner/archive/2007/11/26/dealing-with-memory-pressure-problems-in-moss-wss.aspx
- 84. Memory leak checking tool : http://blogs.msdn.com/b/rogerla/archive/2009/01/29/automate-sharepoint-dispose-code-reviews-with-spdisposecheck.aspx
- 85. Memory Explained; The restaurant analogy: http://blogs.msdn.com/b/tess/archive/2006/09/06/net-memory-usage-a-restaurant-analogy.aspx
Editor's Notes
- #4: One of the remarkable features of .NET development has been the advent of the garbage collector and automatic management of the memory used by programs. This is a tremendous time saver as the developer no longer has to create the de-constructors that were traditionally written in languages such as C++. While this feature is a powerful new tool in the developer’s arsenal, it can also prove to be a dangerous flaw, as numerous programmers have found when working with SharePoint 2007.
- #19: The above code is now threadsafe and caches to a DataTable instead of using an SPWeb object