Performance Data Collection - HUR Journaling Issues
Content

Objective

  • How to collect performance data when there is an issue with Hitachi Universal Replicator (HUR) journaling

Environment

  • Universal Storage Platform V (USP V), RAID600
  • Universal Storage Platform VM (USP VM), RAID600
  • Virtual Storage Platform (VSP), RAID700
  • Hitachi Unified Storage VM (HUS VM), HM700
  • Virtual Storage Platform (VSP G1000), RAID800
  • Virtual Storage Platform (VSP G5x00), RAID900
  • Virtual Storage Platform (VSP Gx00), HM800
  • Virtual Storage Platform (VSP Gxx0), HM850
  • Virtual Storage Platform (VSP) E Series

Procedure

HUR is a pull technology and journaling performance is dependent on a many parts such as:

  • Primary Storage resource performance (MCU)
  • Remote Secondary Storage resource performance (RCU)
  • Replication link performance (for example Network connections, Fabric ISL, WAN, Telecoms)
  • Local replication targets with delta such as TrueCopy (TC) or Global Active Device (GAD) storage
    • Note some configurations do not include a Local or HUR delta system or link

Below is an example of a GAD 3DC HUR solution:

090240412186132.png

It is critical for any Journal diagnosis and investigation that we receive the receive full solution details and coordinated data from all parts as follows:

  1. Please ensure monitoring is enabled at 1-minute intervals and all active Control Units (CU) have been enabled for monitoring on all Storage systems, in particular the Journal Volume CU must be enabled:
  2. Note that detailed performance data is only stored for up to 24 hours so data collection may need careful timing to capture an issue (for example capturing a period of journal growth)
    • There is no reason to provide any data which does not capture a problem or event as we will be unable to progress any investigation
    • On busy systems logs will wrap quickly so please collect data as soon as possible or within 24 hours after an incident
  3. Start mode31 or detailed dumps on all Storage Systems:
  4. Collect performance export data from all Storage systems, RCU, MCU and optional Local Delta (if it is used):
  5. If RPO is rising to unacceptable levels we will require RPO reports for the same period as the performance data
    • Note: we cannot extract RPO from performance or dump data so your script and calculations need to be supplied.
    • Hitachi Protection Manager (part of Ops Centre) can provide RPO reporting
  6. While the data collections complete please add the following information to the case to assist with investigations:
    • Provide details of the issue and the business or application impact
    • Provide a by minute timeline of the issues: first & last occurrence, start/end time of captured problem
    • Identify which Master Journal (MJNL) and/or Remote Journal (RJNL) are having issues
    • Provide an overall design of the MCU, RCU, Local Storage connectivity and any cascaded replication
    • Provide distances and details of replication links including: distances of each link, bandwidth, protocols and equipment used
    • Provide details and schedules of any related "in-system replication" operations using the HUR Primary Volumes (PVOLs) or Secondary Volumes (SVOLs)

Additional Notes

HUR is a pull technology:

  • PVOL write data is packaged up and stored in the Master MJNL which could be in cache or even destaged to JNL-Vol.
  • The Remote RCU Storage Initiator ports will pull data from the Primary MCU Storage target ports.
  • In some models the replication ports may be universal ports capable of both initiator and target roles.
  • The pulling process is know as "journal read" because the RCU reads data from the MCU.
  • The RJNL will make "journal reads" to the Master Journal MJNL and then the RJNL will "restore" data to the target SVOLs.
  • Note that depending on resources part of the MJNL/RJNL may be in cache or on disk in Journal Volumes (JNL-Vols).

Training Videos are available on the Hitachi Vantara YouTube channel:

In-system replication can be a performance factor in a HUR solution especially when used with the UR PVOLs or SVOLs.  In-system replication types are:

  • Shadow Image (SI)
  • Thin Image (TI)
  • Snapshot (COW)
  • Flash Copy (FC) 
Attachments
CXone Metadata

Tags: RAID Data Collection,Data Collection,Universal Replicator,HUR,export,UR,journal,Enterprise Data Collection,collection,journal full

PageID: 138865