What is the concept of ghosting and un-ghosting in SharePoint?

In SharePoint, most site pages derive from templates. The custom pages only store the differences between them. The template is loaded in memory and applied to the custom pages on the fly. In other words, the template is stored in a cache. This definitely brings in performance and flexibility. Flexibility is in terms that when we change the template page, it’s applied to all custom pages. These pages are loaded from the file system. So pages which are loaded from the file system are termed as ghosted pages.

If the page data is loaded from the content database, it’s termed an unghosted page.
As a note, let me clarify the concept of document and content table as we are already trying to understand the concept of ghosting and unghosting. As we know, SharePoint stores all pages in the database. Looking from 50,000 feet, it has two tables: the document table which has the entry of the page and the content which has the source code of the ASPX page.
So when a page is requested, it first checks in the document table and then goes to the content table to load the page. If it does not find the data of the page, it goes to the file directory to load the page. This loading is done by the ASP.NET runtime itself. But if there is data present in the content table, then it’s loaded by the ‘safe mode’ parser.