This topic describes an approach to sizing the Advanced Continuous Historian archive files to meet the data recording needs of a specific implementation. Archives are the files that the Advanced Continuous Historian uses to store historical data. Each archive file contains events for a time period specified by the archive start time and end time. Since the Advanced Continuous Historian uses compression algorithms to reduce the amount of data storage, the information presented here will not provide exact results. However, information to estimate initial archive sizing and a methodology to view the actual data storage rate and adjust accordingly are provided.
The Advanced Continuous Historian system stores data into archive files. You can have multiple archive files of various sizes. The archive files range in size from 2 MB to 2 TB. The default installation configures one file of 256 MB. It is recommended that you create no fewer than three archive files. The maximum number of archives is limited by the operating system and the maximum file handles available, as well as by the available memory in the computer. The theoretical limit is tens of thousands of archive files, although the actual number is typically between 30 to 100 archive files on medium to large systems.
Consider the following when determining archive file size and count:
Take the following items into consideration before setting the archive files:
Follow the general guidelines below to decide how to provision and size the archive files for the Advanced Continuous Historian. For specific questions or requirements, contact Emerson Technical Support.
Consider the items below when deciding on archive size:
Archive size should provide a balance between data rates, storage consumed per unit of time, and shift frequency, number of files per unit of time. Increasing shift frequency is usually better than increasing archive size beyond the recommended range.
You can find data rate information for your archives by using the Archives plugin in the PI System Management Tools (PI SMT), or by running the piartool -al command. Each registered archive shows the Add Rate/Hour which is the average rate of records in kilobytes used per hour to the archive, showing how quickly the archive files are filling up.
Although disk space is usually inexpensive, disk access – in bytes and/or I/O per second – remains expensive. Be sure to apply the appropriate compression settings for tags on large systems as this will reduce disk space used and increase disk access performance. For more information about recommended exception and compression settings, refer to What are recommended Exception and Compression settings (OSIsoft KB 3226OSI8).
Archive throughput will improve when the archive size is significantly smaller than the Windows File System Cache (FSC) so that the most commonly requested data resides in memory. For more information, refer to File Cache Performance and Tuning at the Microsoft TechNet web site. No limit for maximum FSC size exists for Windows Server operating systems.
You can monitor cache and disk performance by looking at the Windows Performance Counters: Cache\Copy Read Hits % and Physical Disk(<Volume>)\% Idle Time. Values significantly above 90% for both of these counters usually indicate a healthy system.
Since the most recent archive files are usually backed up in every system backup, do not create archives that exceed the limits of your physical backup media. Smaller files will also speed up the verification phase of the Advanced Continuous Historical backups.
For specific details about the Advanced Continuous Historian archive sizing and storage space utilization, refer to the Hardware System Sizing Recommendations spreadsheet accompanying the Advanced Continuous Historian documentation.