Open topic with navigation

Disk Drives and RAID Arrays

For increased performance, you should use high performance physical disks such as:

You should also consider distributing the load across multiple disks, for example, using a 4 disk set:

Distributing the load across disks allows for faster access times, as each disk is dedicated to a single purpose. Regardless of the number of disks, you should ensure that each disk has sufficient capacity for its requirements and that your system is set up so that none of the disks become full or near full. For example, the disks used for storing historic or event data while it remains online should be sufficiently large enough to hold the amount of data that is expected to accumulate during that time, while still retaining sufficient free disk space. If you need to retain event or historic data long-term, your system should be set up to archive that data automatically before it reaches the end of its online retention period.

Having sufficient disk capacity—for example, so that disks do not become more that 50% full—also reduces the chances of disk fragmentation.

If your system only has a single disk, you may experience slow performance due to an excessive amount of data needing to be written or read on the disk at the same time. This will depend on the amount of activity on your system and the size of the system as well as the type of disk being used. While partitioning the disk can have some benefits, they are negligible when compared to the effects of adding another physical disk. Partitioning the disk correctly also requires an expert knowledge of partitioning and can be time consuming. Using multiple disks will have a much more noticeable effect on your system performance than partitioning a disk. So if required, you can improve performance by adding separate physical disks. When you have added one or two separate physical disks, you can then move the Historic files and/or Event Journal files to the new disks (see below).

If you can only use a single disk system, you may be able to improve performance by investing in a higher specification disk. For high performance, we recommend Solid State Disks (SSD), rather than mechanical disks.

If your system only has two disks, you need to determine which of the following two directories has the highest level of activity - the Historic directory or the Event Journal directory. To do this:

  1. Access the Server Status Tool (see ‘Working with the Server Status Tool’ in the ClearSCADA Guide to the Server Status Tool) and display the Historic>Historian statistics.
  2. Examine the Read Count and Write Count entries for the Event data and the Raw data:
    • If the total number of Event data Read Count + Write Count entries is significantly higher than the total number of Raw data Read Count + Write Count entries, you should move the Event Journal directory to a separate disk.
    • If the total number of Raw data Read Count + Write Count entries is significantly higher than the total number of Event data Read Count + Write Count entries, you should move the Historic data directory to a separate disk.
    • If the total number of Raw data Read Count + Write Count entries and Event data Read Count + Write Count entries are not significantly different, you can choose whether you want to move Historic data or Event Journal data to another disk.
  3. Change the location of the files by using the System Configuration>Location settings on the Server Configuration Tool (see the ‘Locations of the Database Folders’ section in the ClearSCADA Guide to Server Administration).

With high performance disks or RAID arrays, a multi-disk architecture becomes even more efficient. However, if you cannot implement a high performance system with high performance disks, you can use the same principles mentioned above to take advantage of your existing hardware. By separating a highly active Historic or Event Journal directory (see above) from the other parts of your system, you should be able to improve the disk access speed (and therefore, system performance).

Some file operations in ClearSCADA use database read and write locks. Disk performance can affect ClearSCADA in these cases. You should take this into account when you specify the disks.

ClearSCADA performs the following file operations using a database write lock:

ClearSCADA performs the following file operations using a database read lock:

However, for each of the above, ClearSCADA caches records in memory. Therefore, the disk read operation only occurs if the record is not already in the cache.

Database configuration and data file I/O operations occur without any database locking.

You may notice a negative effect on system performance if:

Disclaimer

ClearSCADA 2017 R2

© 2018 AVEVA Group Plc.  All Rights Reserved.

The Schneider Electric industrial software business and AVEVA have merged to trade as AVEVA Group plc, a UK listed company. The Schneider Electric and Life is On trademarks are owned by Schneider Electric and are being licensed to AVEVA by Schneider Electric.