Common Caches

The following diagram shows the common Jade caches, described later in this section.

Interpreter Cache

Every user (Jade process) has an Interpreter method cache dedicated to it, assuming that multiple caches are being used. Within this cache, methods contend for space. If the user performs functions requiring more methods than can fit in the cache, the least‑recently used methods are removed until there is room to load the new method.

JOM Persistent Object Cache

Every Jade node has a Jade Object Manager (JOM) persistent object cache. Every Jade process running in the node contends for this cache.

Persistent Database Disk Cache

Starting with Jade 7.0, every Jade database has a Persistent Database (PDB) disk cache in the database server. All Jade nodes and server applications contend for this disk cache. All objects that are read in from the database on disk are read through this cache. All persistent objects that are created or updated are written through disk cache. The PDB module uses direct IO to read and write database blocks.

The PDB disk cache is separate from the server node JOM persistent object cache. The disk cache is used only by the PDB itself. The server node JOM persistent object cache is shared by server applications and serverExecution methods, similar to the way that the JOM persistent object cache in any node is shared by all processes running in that node.

File‑System Cache

Jade 7.0 and higher use file-system cache for some things such as non‑database files, but it is not as important as it was in Jade 6.3. Each instance of the Windows operating system maintains a file‑system cache. All Jade databases and other programs running on the server machine contend for space in this cache.

In Jade 6.3, file-system cache was the primary cache for persistent database objects. From Jade 7.0, the primary cache is the PDB disk cache.

Disk Hardware Cache

Most disk subsystems have a built‑in hardware cache. All programs and databases using files on that subsystem contend for space in this cache. If the disk subsystem is shared by multiple machines, then all programs on multiple machines may be contending for this cache. The different machines may be running different operating systems, but they still contend for the same cache.

JOM Transient Object Caches

There is also a JOM transient object cache in every Jade node. All processes in the node contend for it. To support serverExecution methods, there is a remote transient cache in the server node.