JPS6125264A - Information processor - Google Patents

Abstract

PURPOSE:To reduce the overhead that is produced after the rise of a system by loading data to an extended memory according to the information set previously as a part of the system rise operations. CONSTITUTION:A main memory 11, an extended memory 12, an input/output processor 13 and a service processor 14 are connected to an arithmetic processor 10. While external memories 15-17 are connected to the processor 13, and a memory 18 is connected to the processor 14. The data are loaded to the memory 12 according to the information set previously as a part of the system rise operations. This reduces the overhead that is produced after the rise of the system.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a memory control method for a data processing device;
In particular, it relates to methods for loading data into memory.

[Background of the invention]

In a data processing device that uses a virtual memory method, the address space (virtual address space) visible to the software is independent of the capacity of the hardware's real storage device. Furthermore, recently, architectures have become available that can reference virtual addresses of 2 gigabytes (GB).

FIG. 1 shows a conceptual diagram of a data processing device that uses a conventional virtual storage method. In FIG. 1, 1 is an arithmetic processing unit, 2 is a main memory (real memory), and 3 and 4 are external storage devices. The storage device has a hierarchical structure with a main storage device 2 and external storage devices 5 and 40. Programs and data required by the arithmetic processing unit 1 are imported from the external storage device 3.4 to the main storage device 2, and then The data is taken into the processing device 1.

In addition, as information processing systems using TSS y#a terminals become widespread and the number of terminals increases, the amount of data transferred between the main storage device and external storage device increases, and the overhead due to data transfer increases This is becoming the bottleneck for the whole thing. Especially when searching the database from the terminal,
Due to the following reasons, the performance deteriorates significantly and becomes a major problem.

In recent years, databases that are easy to use and suitable for end users have become popular, and data structures have ranged from structured types that are premised on conventional routine tasks to unstructured types that can also be used for non-routine makeshift tasks (user-friendly). (looks like a table).

In these databases, the scope of data searches has become wider than in the past, and the frequency with which databases existing on external storage devices are read into main storage devices has increased, creating a major performance bottleneck.

On the other hand, semiconductor technology has made remarkable progress, and in recent years, large-capacity memories have become relatively inexpensive and it has become easy to increase the capacity of main storage devices. Against this background, the method to solve the above problem is to expand the main memory to a capacity that exceeds the address range that can be accessed by a normal program, and use this expanded area as part of the external storage device. is possible. Furthermore, instead of expanding the main memory device, it may be possible to consider providing a semiconductor device located between the main memory device and the external memory device and using this as part of the external memory device.

(Hereinafter, such a device will be referred to as an extended storage device.) Although the above method reduces the amount of data transferred between the external storage device and the main storage device, it has the following problems.

(1) Expanded storage devices are generally volatile devices such as RAM, and their contents disappear when the system power is turned off.

(2) The capacity of an expanded storage device is generally larger than that of a main storage device, and it takes time to load data. For example, when loading data from a disk with a data transfer capacity of 5MB/8 to 5 (13) expansion storage devices, a minimum of 1,000 seconds is required, and in addition to this, overhead such as disk rotation seek time etc. Time is added.

In this way, even when the system boots up and the database program starts running, it takes a long time for the necessary data to be loaded into the extended storage and data transfer between the external storage and main storage to be reduced. This means that the effects of installing an external storage device will be delayed for a long time.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide an information processing device that immediately produces the effects of an expanded storage device after the system is started up in a system provided with an expanded storage device.

[Summary of the invention]

A feature of the present invention is that in a system equipped with an expanded storage device, data is loaded onto the expanded storage device based on information set in advance as part of system startup. It is characterized by minimizing the overhead.

[Embodiments of the invention]

FIG. 5 is a system configuration diagram of an embodiment of the present invention.

In FIG. 3, 10 is an arithmetic processing unit, 11 is a main memory (MS), and 12 is an extended storage device.

13 is input/output processing unit t (IOP)% 14 is service processor + (8VP), 15 color 17 is IO
Pic Connected external storage device (disk), 1
8 is a storage device (disk/floppy disk) connected to the 8VP vc.

FIG. 2 shows the logical structure (structure visible to the user) of the target database in one embodiment of the present invention. This is represented as a table, and if you think of it in the traditional file concept, a table is almost a file.
Rows correspond to records, and columns correspond to fields within the records. A name (ID) is attached to the table.

FIG. 5, FIG. 8, and FIG. 9 show the flow of processing when implementing the present invention, and will be explained in order below.

The process flow in FIG. 5 shows the process of loading data into the extended storage device based on information from 8VP at 2:00 when the system is started.

Power-on operation when lifting the system (step 60)
After that, an IPL operation (initial program load operation) is performed. (Step 31) During the IPL operation, a command is issued to the SVP. (Step 62) The format of this instruction is shown in FIG. When this instruction 20 is executed, the SvP extracts necessary information from the file device 1B connected to the SvP, as shown in the configuration of FIG.
Write. The contents to be written consist of a control information section and a data section, and include the following contents.

Control information = (Example) (1) The contents of the data section are bits that indicate which format is table ID/physical address format.

(2) Data 1m/) capacity (number of bytes) (3) Capacity of extended area (number of bytes) (4) Start address of the extended area to be loaded.

(There are many ways to express the address.) Data section = (Example) 0) Check whether the inserted data section is a table ID expression from the column control information of the table ID. (Step 35) In the case of table ID expression, the disk 15 where database management information is stored
Load database management information from. (Determine the total capacity of the table to be step-loaded from the database management information. (Step 55) Check whether there is enough free space in the expanded storage device. (Step 36) If there is a exhausted area, also read the management information. Toni.

Load the necessary data and update the necessary tables. (
Step 57). One embodiment of loading may be done using, for example, a B-tree format index as shown in FIG.

In the example in Figure 6, the index is divided into 6 levels, and the actual data (records) are on pages 55 and 54.
It's inside. This page is generally 4 KB, and in this example, data is loaded into the extended storage device in units of pages.

For example, the data 55 is indexed by the index 51 which is also pointed to by the topmost index 50, and the page 53 containing the data 55 by the index 52 which is pointed from there.
You can get the following. In addition to the data pages described above, an index as shown in FIG. 6 may also be loaded into the extended area. Such pointers are generally addresses on the disk (addresses on the database), and after loading them into the extended area, it is necessary to create a correspondence table. The table shown in FIG. 7 is such that a pair is created using the address 61 on the database and the address 62 directly above the loaded extension notation, and when performing an actual search, the extended storage device address VC is converted while looking at this table, and the data is will be referred to.

Check whether there is a table ID to be loaded.

(Step 38) If there is, the process branches to step 35 and the above-mentioned process is repeated. If not, normal OS loading operation is started (step 59).

In addition, if a physical address is specified in step 53, the physical address of the data section (expression such as CCHHR)
Load data based on.

Figures 8 and 9 show the processing from when the database actually starts working and the user starts using it, to deciding which table D should be loaded the next time the system is started up, to notifying the SVP tic. . There are various ways to select table IDs, but in this embodiment, the two methods shown in FIG. This method creates tables until the database is reached. First, when a user (terminal) requests access to the database, the table ID is obtained. (Step 70)
Check whether the table has been registered in the registration table 78. (Step 71) If it has been registered, return to step 70 at the entrance; if it has not been registered, register the table ID, update the pointer 77, and change the number of pages related to the car crest table IDVc to the total number of pages 79. Add. (The initial value of the total number of pages is zero) (Step 72
) Check whether the total number of pages 79 exceeds the number of pages in the expanded storage device. (Step 74) If it has not been exceeded, return to step 70 at the entrance; if it has been exceeded, necessary control information is added, (Step 75)
An instruction is issued to the 8VP vc using an instruction similar to the instruction format shown in FIG. 4 for control information and table ID columns. (Step 76) FIG. 9 shows a method in which a certain period of time is set, and when that period of time has elapsed, a table ID column is created based on the table ID string that appears frequently in the certain period of time.

First, set the timer. (Step 80) Thereafter, when a user (terminal) requests access to the database, the table ID is obtained. (Step 81) Check whether the table ID has been registered in the needle bull 88.

(Step 82) If it has been registered, skip step 85 and proceed to step 84. If it is not registered, register the table ID in registered table 8BVC. (Step 86) After registration, the count 89 corresponding to the table ID is updated (+1). (The initial value of the count 89 is zero eight (step 84), and the process returns to step 81K and repeats the above processing. After a certain period of time has elapsed, a timer interrupt 99 occurs, and the count 89 of the registered table 88
Based on the value of , sort in descending order of number of requests. (Step 85) Then, calculate the required number of pages from the upper page ID, add enough tables and control tables to fit into the expanded storage device, (Step 86) control information and tables,
The ID number 11 is set to S by an instruction similar to the instruction format shown in FIG.
Issue orders to VP. (Step 87) In this embodiment, data load information is held in 8VP, but it is not necessarily held in 8VP, and may be held on a disk via a general channel.

〔Effect of the invention〕

As is clear from the above explanation, the original F! According to AK, data can be loaded onto the expanded storage device based on preset information as part of starting up the system, so the overhead caused by the expanded storage device after the system startup can be minimized. This makes it possible to quickly respond to requests from terminals even after the system is set up.

[Brief explanation of the drawing]

Fig. 1 is a conceptual diagram of a conventional data processing device, and Fig. 2 is a conceptual diagram of a conventional data processing device.
Logical structure diagram of the database subject to the present invention, third
The figure is a system configuration diagram of an embodiment of the present invention, and FIG.
FIG. 5, FIG. 8, and FIG. 9 are process flow diagrams of an embodiment of the present invention. FIG. 6 is a diagram showing an example of the instructions used in the present invention. FIG. FIG. 7 is a database-related address correspondence diagram. 10... Arithmetic processing unit 11... Main storage device 12.
...Extended storage device 15...I/O processing device 14...
・Service processor 15-18...external storage device

Claims (1)

[Claims] 1. In a system in which a large-capacity storage device is set between a main storage device and an external storage device, a first means for holding information regarding the initial data load of the storage device; a second means for loading data into;
an information processing apparatus comprising a third means for creating information regarding initial data loading, and loading data into the storage device based on preset information at the time of system startup.