Browse Prior Art Database

System and method to implement host tiered hot spot tuning on storage

IP.com Disclosure Number: IPCOM000202012D
Publication Date: 2010-Dec-01
Document File: 6 page(s) / 105K

Publishing Venue

The IP.com Prior Art Database

Abstract

This paper is to take host side tiers into the consideration when doing hot spot tuning at storage device side. It enables the user to customize the prioritization process over the hot spot management plan in a resource limited storage system. The basic idea is to differentiate the hot spots both in analysis and in management regarding their significance, so as to balance between the performance requirement and the resource limitation.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 42% of the total text.

Page 01 of 6

System and method to implement host tiered hot spot tuning on storage

1.Background:

In some storage devices, hot spot management is a new and useful feature to optimize the usage of users on backend volumes. The key implementation is:
storage controller will collect I/O performance statistics on volumes;
there is a criteria to decide whether or not there is hot spot on a specific volume;
if there is hot spot on the volume, relocation will be done to move the volume from slow disk (etc. HDD) to fast disk (etc. SSD - Solid State Disk).

But there are some problems on the current function support:

Current smart hot spot tuning are using the same rule (on/off) to all host sides, so

Host side privilege service is broken

     
Part of host side ILM service is broken, if all disks are within one storage
Current smart hot spot tuning uses the same rule to decide whether or not to relocate volume, but some factors should be handled differently according to different requirement:

Based on I/O request number


Based on I/O request size (useful to multi-media service vendor)

This paper is to implement different hot spot tuning policies at different levels within one storage system, and hot spot tuning action can be taken according to different policies on different levels. The different levels can be:

whole system, storage pool set, each storage pool, volume set, each volume.

Here is the solution diagram:

Host 1

Host n

Storage Controller(s)

Old I/O

New I/O

New Volume

Old Volume

Old extents

Slow HDDs

New extents

Fast SSDs

1


Page 02 of 6

(This page contains 00 pictures or other non-text object)

2.Solution summary:

This paper is to take host side tiers into the consideration when doing hot spot tuning at storage device side. It enables the user to customize the prioritization process over the hot spot management plan in a resource limited storage system. The basic idea is to differentiate the hot spots both in analysis and in management regarding their significance, so as to balance between the performance requirement and the resource limitation.

The overall solution is as follows:

Storage Controllers

Hot spot measurement Policy

Policy Module

Policy based Hot spot

Execution Module

Hot spot tuning Policy

Policy Store

Storage Pools

Storage Volumes

Low speed disks

The main idea is comprised of three steps:
a. Classify the storage entity into priority-based tiers for host
In this paper, the user classifies the storage system into multiple tiers per different user priority or per host privilege. The granularity can range from storage pools, LUNs and down to sub-LUNs.


b. Customize the hot spot measures for each tier

2

Medium speed disks High speed disks


Page 03 of 6

Instead of having a universal formula for defining a hot spot,

in each tier to measure the level of "heat" based on multiple performance statistic factors (See the N measurements in Fig.1). The performance statistics factors could include I/O number, I/O request number, I/O response time, etc. In th...