Browse Prior Art Database

DISK SYNCHRONIZING USING. FIXED RECORD COUNTS

IP.com Disclosure Number: IPCOM000006807D
Original Publication Date: 1993-Mar-01
Included in the Prior Art Database: 2002-Feb-04
Document File: 3 page(s) / 139K

Publishing Venue

Motorola

Related People

Bradley L. Frohman: AUTHOR

Abstract

Many systems in the telephone industry are now redundant systems. A redundant system is defined as any system that offers no single point of failure. The reason for redundant systems is to try to ensure that data and service are not lost. The problem that com- monly arises in redundant systems is how to keep the disks synchronized. There are certainly a number of ways to try to produce duplicate copies of data on disks to try to insure data is not lost. For the sake of this article it is assumed the redundant system must store to a hard disk (as opposed to memory or tape).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 47% of the total text.

Page 1 of 3

MO-LA INC. Technical Developments Volume 18 March 1993

DISK SYNCHRONIZING USING. FIXED RECORD COUNTS

by Bradley L. Frohman

   Many systems in the telephone industry are now redundant systems. A redundant system is defined as any system that offers no single point of failure. The reason for redundant systems is to try to ensure that data and service are not lost. The problem that com- monly arises in redundant systems is how to keep the disks synchronized. There are certainly a number of ways to try to produce duplicate copies of data on disks to try to insure data is not lost. For the sake of this article it is assumed the redundant system must store to a hard disk (as opposed to memory or tape).

  There are a number of types of solutions that can be tried in trying to keep disks on redundant system to be synchronized. One disk solution is to have mirrored disks. This offers duplicate copies. It does not offer "no single point of failure:' Since disks are often the most likely part to fail, the most desirable telephone system is one that is duplex yet completely redundant and inde- pendent in parts and information. The problem to solve on this type system is how to guarantee that the infor- mation on each subunit in the system is identical. More specifically it would be advantageous for maintenance purposes if the information was identical at the tile level. This would make failure recovery much simpler. In the systems the author has seen, this duplication of data was always done using a time period For example every hour the current tile was closed and a new file was opened. The problem with this solution in a redundant system is that since there are two clock generators, the files may not be identical as the clocks drift. (There must be two clock generators, or there is a single point of failure).

  The proposed idea is to use futed record counts to determine when to close and re-open the files (instead of time periods). The purpose of this idea is to provide not only logically equivalent storage on a redundant sys- tem, but to provide physically equivalent storage. For the sake of this article I have chosen a store and forward scenario. In this scenario a transmitting device will send data to a redundant system (See Figure 2). The redun- dant system will have to be a multi-tasking environment

148

to allow for data receiving gnd file copying (after failure) to occur at the same time. ;In the figure shown a "Data Transmitter" sends the da& to be collected. This data is then split. Typically this data would be split by using a Y-cable. Each sub-unit w'ithin the redundant system receives the input data and partitions the data based on a record count. By this means each output block (X, Y,. .) will be identical. This in turn implies each record (n,n+l,. .m,m+l,. .z) in e&h block will be identical.

  In order to confxm this as a valid solution it is nec- essary to show how all the normal and abnormal cases will be handled. (see Figure 1).

1....