Showing posts with label DMX. Show all posts
Showing posts with label DMX. Show all posts

Symmetrix Optimizer improves array performance by continuously monitoring access patterns and migrating devices to achieve balance across the disks in the array. This is process is carried out automatically based on user-defined parameters and is completely transparent to end users, hosts and applications in the environment. Migration is performed with constant data availability and consistent protection.

Optimizer performs self-tuning of Symmetrix data configurations from the Symmetrix service processor by:
· Analyzing statistics about Symmetrix logical device activity.
· Determining which logical devices should have their physical locations swapped to enhance Symmetrix performance.
· Swapping logical devices and their data using internal Dynamic Reallocation Volumes (DRVs) to hold customer data while reconfiguring the system (on a device-to-device basis).

Symmetrix Optimizer can be utilized via EMC Symmetrix Management Console or SYMCLI, where user can defines the following:

1) Symmetrix device to be optimized
2) Priority of those devices.
3) Window of time that profiles the business workload.
4) Window of time in which Optimizer is allowed to swap.
5) Additional business rules.
6) The pace of the Symmetrix Optimizer volume copy mechanism.

After being initialized with the user-defined parameters, Symmetrix Optimizer operates totally autonomously on the Symmetrix service processor to perform the following steps.

1) Symmetrix Optimizer builds a database of device activity statistics on the Symmetrix back end.

2) Using the data collected, configuration information and user-defined parameters, the Optimizer algorithm identifies busy and idle devices and their locations on the physical drives. The algorithm tries to minimize average disk service time by balancing I/O activity across physical disk by locating busy devices close to each other on the same disk, and by locating busy devices on faster areas of the disks. This is done by taking into account the speed of the disk, the disk geometry and the actuator speed.

3) Once a solution for load balancing has been developed the next phase to carry out the Symmetrix device swaps. This is don using established EMC Timefinder technology, which maintains data protection and availability. Users can specify if swaps should occur in completely automated fashion or if the user is required to approve Symmetrix device swaps before the action is taken.

4) Once the swap function is complete, Symmetrix Optimizer continues data analysis for the next swap.

How Symmetrix Optimizer works:-

1) Automatically collects logical device activity data, based upon the devices and time window you define.

2) Identifies “hot” and “cold” logical devices, and determines on which physical drives they reside.

3) Compares physical drive performance characteristics, such as spindle speed, head actuator speed, and drive geometry.

4) Determines which logical device swaps would reduce physical drive contention and minimize average disk service times.

5) Using the Optimizer Swap Wizard, swaps logical devices to balance activity across the back end of the Symmetrix array.

Optimizer is designed to run automatically in the background, analyzing performance in the performance time windows you specify and performing swaps in the swap time windows you specify.

Veritas Disk Group Configuration Guidelines:-

1) Use multiple Disk Groups—preferably a minimum of four; place the DATA, REDO, TEMP, UNDO, and FRA archive logs in different (separate) Veritas Disk Groups

2) Optimally, use RAID 1 for tier 1 storage

3) Configure Disk Groups so that each contains LUNs of the same size and performance characteristics

Distribute Veritas Disk Group members over as many spindles as is practical for the site’s configuration and operational needs

Data Striping and Load Balancing:-

1) Veritas software level striping: layout=stripe ncols=10 stripeunit=128k

2) Storage-level striping further parallelizes the individual I/O requests within storage

3) Using the storage RAID protection, the amount of I/O traffic (host to storage) is reduced

4) EMC PowerPath should be used for load balancing and path failover

5) Use of metavolumes is optional

a) There is an upper limit on the number of LUNs that a host can address—typically ranging from 256 to 1,024 per HBA.

b) When these limits are reached, metavolumes are a convenient way to access more Symmetrix hypervolume.


Volume Configuration with Veritas (Hypervolumes):


1) Created 5 Veritas Disk Groups

2) Five Disk Groups are used because this number provides better granularity for performance planning

3) The use of five Disk Groups also provides increased flexibility when planning for the utilization of EMC replication technology within the context of an enterprise-scale workload

4) Having five Disk Groups permits the placement of data onto different storage tiers if desired

Hypervolume

Purpose

Size

1

DATA

32 GB

2

REDO

400 MB

3

DATA

32 GB

4

FRA

30 GB

5

TEMP

10 GB

6

FRA

30 GB

Average Disk Utilization for Raid 1 should be below 150 IOPS per disk and should not go above 200 IOPS per disk as per below configuration.

-- 80 physical disks (40 mirrored pairs)

-- 240 devices visible to Veritas

-- Average user count ~ 16,000

About Me

My photo
Sr. Solutions Architect; Expertise: - Cloud Design & Architect - Data Center Consolidation - DC/Storage Virtualization - Technology Refresh - Data Migration - SAN Refresh - Data Center Architecture More info:- diwakar@emcstorageinfo.com
Blog Disclaimer: “The opinions expressed here are my personal opinions. Content published here is not read or approved in advance by EMC and does not necessarily reflect the views and opinions of EMC.”
EMC Storage Product Knowledge Sharing