Performance Degradation issues in Exalogic X2-2 Racks when ZFS 7320 Appliance configured with 24GB RAM

Asset ID:	1-79-1967979.1
Update Date:	2015-02-10
Keywords:

Solution Type Predictive Self-Healing Sure

Solution 1967979.1 : Performance Degradation issues in Exalogic X2-2 Racks when ZFS 7320 Appliance configured with 24GB RAM

Applies to:

Oracle Exalogic Elastic Cloud X2-2 Hardware
Oracle Exalogic Elastic Cloud Software
Oracle Solaris on x86-64 (64-bit)
Linux x86-64
Oracle Virtual Server (x86-64)

Purpose

In-rack ZFS 7320 Appliance with 24GB RAM on Older Exalogic X2-2 Racks May Cause Performance Degradation.

Exalogic X2-2 with ZFSSA 7320 configured with 24G RAM per storage control head. To check the amount of RAM currently installed, perform the following steps:

Login to the ZFS BUI
Navigate to Maintenance -> HARDWARE
View the installed memory as shown in the screenshot below:

ZFS

Alternatively, run Exachk and from the generated Exachk report, review the ZFS Storage Appliance Check for installed RAM size.

Details

Symptoms

The ZFS 7320 Storage Appliance on older Exalogic X2-2 racks were configured with 24GB RAM per storage control head. Customers who have deployed such racks may experience mild to severe performance issues for the following workloads:

Exalogic rack configured in the D/R site hosting a virtual data center (vDC) with active replication streams from the primary site
iSCSI used to access ZFSSA for the following:
- E-Business Suite deployment on Bare Metal Linux
- Solaris Zones (Zones over Shared Storage)
In-rack ZFSSA used to host database backup (Note: this is not recommended on Exalogic)
Exachk reports failed check described in Exachk Reports: Document ID 1682174.1 - L2ARC Header Size Exceeds the Recommended Limit

Solution

Upgrade the ZFSSA control head from 24G RAM to 96G RAM per head. Contact Advanced Customer Service (ACS) for Oracle Exalogic, or your Fusion Middleware Sales Manager for details on acquiring the memory upgrade kit.

Follow the standard Memory replacement service procedure with the following requirements specific to Exalogic:

Always start with the passive head
Leverage Planned Graceless recovery feature by using one of the following methods to power down the head
- Login to the BUI and click the power icon on the left side of the masthead
- SSH into the storage system and issue the maintenance system poweroff command
DO NOT use failback

Attachments

This solution has no attachment