Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
SRM Scenarios
The following scenarios describe projects in which SRM systems improved the efficiency of the administration and planning of large, distributed computing storage infrastructures. These scenarios are not comprehensive outlines of storage-related efforts. They are descriptions of efforts in which SRM systems are used to increase the operational efficiency and management of distributed system storage resources. These systems have empowered the existing storage administration staffs to take on greater responsibility without adding head count by reducing the effort of administration through the adherence to SRM best practices. These scenarios hopefully shed some light on how to use storage resource management systems and on the best practices for such systems to increase the efficiency and management of distributed systems storage infrastructures.
The scenarios are presented in two parts:
Strategic scenarios focused on planning for large infrastructure efforts
Tactically aligned scenarios that focus on consumption management
All of the scenarios are defined within the context of the following key SRM functions.
TABLE 1 Key SRM Functions
|
SRM Functions |
Description |
|
Storage discovery |
Finds the detailed information about enterprise storage resources that can be used to better manage the enterprise storage stack. |
|
Capacity planning |
Projects capacity growth for one or more storage objects under SRM management. |
|
Consumption management |
Assists storage customers and administrators to mediate capacity consumption rates. |
|
Data migration |
Assists in the identification and planning stages of data migration efforts. |
|
Charge back |
Facilitates the development of accurate charge-back systems based on customer utilization habits. |
NOTE
These key SRM functions are not mutually exclusive, but complementary. See "Appendix" on page 23 for an outline of the basic cost-avoidance calculations for the first scenario. Use this appendix to document return-on-investment (ROI) from the implementation of an SRM system.
SRM Strategic Scenarios
This section contains the scenarios that show the strategic value of SRM practices.
Scenario 1: Corporation W
Key SRM Functions: Storage Discovery – Capacity Planning
In this scenario, SRM helps to discover significant storage-related trends resulting in increased accuracy of the capacity planning decision-making process.
Corporation W is preparing to make strategic budgeting decisions for the upcoming year. One of these decisions is whether or not to continue expanding their enterprise storage capacity for distributed systems. Last year they expanded their UNIX_ and NT system enterprise storage capacity by 100%, and this year the businesses are planning a similar expansion. However, before agreeing to absorb the cost of yet another 100% expansion, senior management realizes that it needs better visibility within the current enterprise storage stack to calculate the actual data growth rate.
The question is simple: Based on actual data utilization growth metrics, can we see a real need to grow our current capacity by 100%? The enterprise storage operational staff must answer three basic questions:
What is our current enterprise storage capacity?
What is our current enterprise storage utilization?
What is the growth rate of the enterprise storage stack?
No current reporting mechanism can offer this visibility. They have no way to trend capacity and utilization across all systems. They decide to implement an SRM system to provide them with the necessary data. An SRM product is selected and implemented on UNIX and NT servers within the firm. A management overview of the firm's storage is developed for senior management. The following data shows the findings from the SRM storage discovery phase.
TABLE 2 SRM Storage Discovery-Capacity Planning Table
|
Overall Capacity |
Free Space |
Used Space |
Percent Utilized |
Quarterly Growth Rates |
|
12 Tbytes |
5 Tbytes |
7 Tbytes |
58.33% |
12% |
A quick analysis of this data produces an annual growth chart, as follows.
){kind=small}
FIGURE
1 SRM Capacity Planning Chart
Based on the actual recorded growth rate of 12% per quarter (48% annual increase), the enterprise storage stack within the year requires only an additional 6 Tbytes of capacity to maintain existing utilization rates (58.33%). The pre-SRM estimated capacity requirement of an additional 12 Tbytes (100% expansion or 25% quarterly growth) would have meant the purchase of 6 Tbytes of unnecessary disk. Assuming a cost-per-megabyte of $0.302, 6 Tbytes of unnecessary disk, or 6,000,000 Mbytes, would cost the firm $1,800,000.00.
In this scenario, SRM best practices are used to assist senior management in understanding what storage costs are actually warranted based on efficient storage trends monitored within the enterprise storage stack. This view, which is strategic in scope, is one area where SRM systems can add significant value to enterprise storage organizations.
Scenario 2: Corporation X
Key SRM Functions: Storage Discovery – Capacity Planning – Data Migration
In this scenario, SRM discovers significant storage-related trends within logically configured storage objects. This information is then used for more efficient SAN planning.
A division within Corporation X is considering a SAN to replace its current stock of locally attached storage resources. The primary drivers for this move are:
Utilization levels of locally maintained file systems are inefficient.
Administrator-to-terabyte ratios are high due to the numerous and geographically dispersed local storage resources.
Backup and recovery windows are becoming harder to maintain for the numerous and geographically dispersed local storage resources.
As part of their pre-SAN planning, Corporation X implements an SRM system to ascertain what data to target for SAN migration and how behavioral characteristics of this data might impact SAN configuration. They must answer three key questions before planning the SAN size and configuration.
TABLE 3 SRM SAN Planning Table
|
SRM-Related Question |
SAN Planning Impact |
|
What was the current size of the data to be migrated to the SAN (UNIX___–NT servers)? |
Initial capacity of the SAN and SAN zone planning |
|
What is the annual growth rate of the data to be migrated (UNIX___–NT servers)? |
Initial SAN capacity requirements and projected SAN capacity requirements after one year |
|
What are the characteristics of the data and the data activity levels (modification frequency on UNIX–NT servers)? |
SAN backup planning and prioritization of data migration to the SAN |
The answers to these questions determine the size, growth rate, and characteristics of the data that is deemed both critical and dynamic enough to be migrated to a new SAN.
Key personnel from the storage organization and divisional business units are brought together to construct the logical groupings of divisional storage resources by data function, as shown in the following table.
TABLE 4 Logical Groupings of Divisional Storage
|
Logical Groupings |
Function |
|
Application data |
Production applications |
|
User data |
User community's personal and/or business needs |
|
Database data |
Database file repositories |
To assist in understanding how different data types behave, these logical groupings are then monitored and trended by the SRM system.
These groups comprise:
- Server groups by OS type
- File system groups by OS type
- Directory or folder groups by OS type
Based on these groupings, they developed the following SAN capacity planning chart.
){kind=small}
FIGURE
2 SRM SAN Capacity Planning Chart
Both the user and database data types are the most likely targets for migration to the SAN, based on their dynamic growth patterns, as follows:
- User data is projected to grow by 50% within the year.
- Database growth is projected to reach 133% within the year.
- Application data is projected to be tame at 2% within the year.
Having determined that the database and user data are the first data types to be migrated to the SAN, the storage organization and its business customers determined that the initial SAN capacity must be 7.5 Tbytes to accommodate the initial 5 Tbytes of user and database data plus one quarter's worth of projected growth for both data types. This leaves the SAN with a projected 20% capacity buffer by the end of the first quarter. This margin is deemed sufficient to buffer SAN utilization growth while future SAN capacity expansions are put in place. Yearly budgeting for SAN storage includes the cost of 12 Tbytes of SAN storage within the year. This number includes the projected year-end capacity requirements for user and database storage capacity (10 Tbytes) plus a 20% buffer.
The capacities and growth rates by operating system allow the storage administration group to efficiently plan initial zone capacities within the new SAN. The SRM data shows the following data growth by OS type.
){kind=small}
FIGURE
3 SRM Growth Rates by Operating System Type
From this data, storage organization was able to determine the following:
Zone for the UNIX OS must support an initial capacity of 4 Tbytes, with a projected annual growth rate of 75%.
Zone for the NT OS must support an initial capacity of 2 Tbytes, with a projected annual growth rate of 50%.
By applying growth rates by operating-system metrics to their planning stage, the storage administrators can more efficiently plan not only for overall SAN capacity, but also for efficient zone capacity configuration within the SAN.
SRM SAN Data Migration
The SRM metrics for the modification trends within data types are collected to help prioritize the migration of production data onto the SAN. This prioritization also helps ensure that production data that currently requires the most backup and restore efforts within the older, geographically dispersed infrastructure is the first to be moved to the new SAN. The plan is to identify the most dynamic data within each data type for first-phase migration. The capacity of data modified on a nightly basis also helps plan for the capacity of nightly backups within the new SAN. This backup-related data is considered very important because the ability to centralize critical backup and restore operations within the SAN is key to reducing costs while increasing efficiency for enterprise storage within the firm. Therefore, proper sizing and efficient operation of the new SAN backup is central to the success of the migration plan.
The following charts show the modification metrics for user data and database data.
){kind=small}
FIGURE
4 SRM Modification Metrics by Data Type
){kind=small}
FIGURE
5 SRM Data Modification Capacity Charts
Using these modification capacity metrics, and the server, file system, and directory groupings within the SRM system, the storage administrators can map this modification capacity back to the individual directories, file systems, and, ultimately, servers on which this data is currently stored. They can now plan data migration on a server-by-server, file-system-by-file-system basis from the old infrastructure to the new SAN-based infrastructure, based on accurate data criticality and activity information.
The storage administration group can also plan for a SAN backup capacity of at least 5 Tbytes within the first month of operation, based on modification trends within the targeted migration data, as shown in the following modification trends chart.
){kind=small}
FIGURE
6 SRM Data Modification Trends
Target data that consistently shows modification stamps of over a month is scheduled for later migration to the SAN because its backup and restore activity does not pose a burden to the existing storage infrastructure. Data that shows modification stamps of over a year is deemed least important to migrate to the new SAN, and steps are taken to review less expensive storage architectures onto which this largely archival data can be moved.
In this scenario, SRM best practices were used to efficiently plan a new SAN. SRM metrics provide guidance for the initial size of both SAN storage capacity and backup capacity, as well as the most efficient path of data migration from the existing storage infrastructure.
SRM Tactically Aligned Scenarios
This section contains SRM best practices that add value through tactical efforts, such as consumption management.
Scenario 1: Corporation Y
Key SRM Components: Storage Discovery – Capacity Planning – Consumption Management
In this scenario, SRM helps to facilitate consumption management strategies.
Business unit managers within a division of Corporation Y are looking for a better understanding of how their storage is consumed. They want to develop storage best practices that can ultimately lead to a reduction in their storage-related costs, along with improved storage availability.
These unit managers ask the storage administration group to set up metrics within the SRM system to help them make better storage-related decisions. Working with the storage administration group, they create logical SRM groupings representing the individual business units within the division, as well as logical units representing two areas of user-related consumption: personal data and NT user profile directories. After these objects are placed under SRM management, the business unit managers provided a list of data points to the storage administrators to help them determine storage best practices within their areas.
The business-requested SRM user-data-related metrics are:
- Number of users
- Capacity of all user data areas
- Capacity of all user profile areas
- Average consumption by user for personal and profile areas
- Growth rates on all personal user data areas
- Growth rates on all user profile areas
- Growth rates on individual user data directories
- Growth rates on individual user profile directories
- Growth spikes in user personal data areas
- Growth spikes in user profile directories
- Lists of user directories over a designated size
- Lists of user profile directories over a designated size
Using these data points, two different reports are created for the business managers:
The strategic report details the total capacity for user data and profile directories, the average consumption for both these areas, and the growth rates for these two areas across all server resources.
The tactical report lists largest current consumers of space, user data, or profile directories that have grown greater than a designated threshold between daily SRM scans.
Using these two report types, the business managers, in conjunction with their storage administrators, hope to better determine a path to storage best practices for their areas.
TABLE 5 SRM Consumption Management Business Unit Report
|
Unit Number |
Users |
Data Capacity |
Profile Capacity |
Average Data Capacity |
Average Profile Capacity |
|
Unit 1 |
117 |
53,259 |
10,652 |
455 |
91 |
|
Unit 2 |
154 |
40,437 |
8,087 |
263 |
53 |
|
Unit 3 |
134 |
28,056 |
5,611 |
209 |
42 |
|
Unit 4 |
86 |
10,661 |
2,132 |
124 |
25 |
|
Unit 5 |
23 |
3,301 |
660 |
144 |
29 |
|
Unit 6 |
12 |
3,352 |
670 |
279 |
56 |
|
Unit 7 |
192 |
44,999 |
9,000 |
234 |
47 |
|
Unit 8 |
193 |
17,804 |
3,561 |
92 |
18 |
|
Unit 9 |
183 |
52,845 |
10,569 |
289 |
58 |
|
Unit 10 |
43 |
12,019 |
2,404 |
280 |
56 |
|
Totals |
1,137 |
266,733 |
53,347 |
235 |
47 |
){kind=small}
FIGURE
7 SRM Consumption Management Charts
){kind=small}
FIGURE
8 SRM User Profile Growth
){kind=small}
FIGURE
9 SRM Average Consumption Trends
An analysis of the SRM consumption management data shows that average user data and profile storage consumption will increase by 54% and 55%, respectively, within the next year if no action is taken to manage consumption.
In response to this consumption analysis, quotas are designated for the data and profile directories for each individual user. These quotas are derived from the SRM average utilization metrics compiled in the SRM consumption management reports. Quotas for the data areas are set to 400 Mbytes, and profile quotas are set to 50 Mbytes. This action gives business managers and storage administrators a year to incorporate the data area quotas into their daily management routines and to establish a hard storage practice that eliminates storage consumption related to the growth of user profile directories.
To support these consumption management efforts, quotas enforced by the SRM system are configured to alert users and their storage administrators when quotas are exceeded, and the tactical SRM reports are published for review by the business unit managers. In this way, considerable storage consumption related to user profile capacity can be avoided, while user data area growth is placed under a much more efficient management model, which clearly delineates user-data-area offenders.
Through the implementation of SRM best practices related to the consumption of logical user data objects (user data and profile directories), storage administrators and business users efficiently implemented storage best practices that beneficially impact costs, administration, and availability of their distributed storage resources.
Scenario 2: Corporation Z
Key SRM Components: Storage Discovery – Capacity Planning – Data Migration – Consumption Management – Charge Back
In this scenario, SRM helps to facilitate consumption management, leading to the development of a storage-utilization-based charge-back system.
Within Corporation Z, storage administrators want to target a specific data class, which they think might be responsible for significant storage consumption. They configure the SRM system to discover and trend consumption related to archive messaging folders within business units. They identify this class of data because archive messaging folders are suspected to:
- Exist on a large number of servers
- Contain large files
- Undergo frequent modifications
- Require frequent backup and restore overheard
They hope that by correctly classifying behavior within this data classification, they can properly plan for archive messaging folder consumption going forward.
TABLE 6 SRM Storage Discovery and Capacity Planning Table
|
Item |
Value |
|
Number of servers housing archives files |
200 |
|
Total capacity |
2.25 Tbytes |
|
Utilization |
25% (of existing utilization) |
|
Quarterly growth |
15% |
|
Projected annual growth |
1.6 Tbytes |
|
Total number of folders |
20,000 |
|
Folders under 100 Mbytes |
40% |
|
Folders between 100-400 Mbytes |
49% |
|
Folders between 400 Mbytes and 1 Gbytes |
9% |
|
Folders over 1 Gbytes |
2% |
|
Number of folders over 500 Mbytes |
1,600 (8%) |
|
Space consumed by folders over 500 Mbytes |
45% (1 Tbytes) |
|
Average messaging folder size |
160 Mbytes |
SRM Data Migration
SRM data indicates that the archive messaging folders are growing quickly over very distributed resources. This finding substantiates the concerns of the storage administration staff. Archive messaging folders are large, active, and dispersed. These archives require a very high degree of maintenance with a large and growing impact on backup and restore operations, as well as general file-system-based storage outages. Only 11% of the archive messaging folders are consuming 45% of all consumption related to archive messaging folders. This indicates that a fairly small community of users is consuming an inappropriately large amount of storage.
The storage administrators decide upon the following actions:
Centralize archive messaging folder capacity.
Place strict quotas on folder size.
Implement a tiered charge-back system to assist storage customers in understanding the costs associated with their storage practices.
In preparation for the mail folder move to centralized repositories, SRM is used to determine the following:
Exact location of each archive messaging folder file
Number of archive messaging folders within each divisional business unit to be migrated
Capacity of all archive messaging folders within each divisional business unit to be migrated
The following table shows the SRM data migration numbers.
TABLE 7 SRM Data Migration Table
|
Business Unit |
Number of Servers |
Percentage of Mail Folder Capacity |
Number of Files |
|
1 |
15 |
25% |
4,000 |
|
2 |
25 |
15% |
3,500 |
|
3 |
50 |
37% |
7,500 |
|
4 |
10 |
23% |
5,000 |
With this data in hand, and with minimal service disruption, the storage administrators now plan an orderly migration of archive messaging folders by business unit on a server-by-server basis to the new centralized archive messaging folder storage.
Storage administrators configure the new SAN-based storage to allow the SRM system to monitor consumption on newly centralized directory structures devoted to archive messaging folders. These structures represent their respective business unit customers and can be tracked as logical units within the SRM system. Administrators then trend these logical units and monitor them for use in a charge-back system.
SRM Consumption Management
The next step for the storage administrators is to properly determine archive messaging folder quotas for use in conjunction with charge back system. They consult the SRM data to better understand folder distribution count (FIGURE 10) and size metrics (FIGURE 11).
){kind=small}
FIGURE
10 SRM Folder Counts by Size
){kind=small}
FIGURE
11 SRM Folder Consumption Management by Size
With this information in hand, the storage administrators determine that 92% of all archive messaging folders is under 300 Mbytes in size with an average size of 160 Mbytes. However, these same folders are responsible for only 33% of the capacity currently consumed by all archive messaging folders. By targeting only 8% of their current archive messaging folder population to conform to a new 300 Mbytes quota policy, they can realize significant storage consumption reductions across 67% of the space currently associated with archive messaging folder consumption. With the SRM system in place, identifying this 8% population is easy and reliable, down to the individual users responsible for the consumption.
SRM Charge Back
Armed with the consumption management data, the storage administrators work closely with their business customers to develop a charge-back system that supports a new agreement tailored to enterprise storage service levels. This agreement can ultimately help their customers control storage consumption and associated costs. The charge-back system is fed by the SRM data. Using the SRM system data, the charge-back system for archive messaging folders tracks for the following items:
- Consumption for the entire division
- Consumption by business unit
- Consumption by user
- Consumption within quota by business unit
- Consumption over quota by business unit
- Percentage of business unit users over quota
- Percentage of business unit users within quota
- Individual users within business units within quota
- Individual users within business units over quota
- Daily, quarterly, and annual growth rates by business unit
This charge-back system, developed through the implementation of SRM best practices, provides a means of identifying behavioral characteristics of a logical storage object (archive messaging folders) active within the physical storage resources of a distributed storage infrastructure. This data enables storage administrators and business users define storage best practices impacting the availability, cost, and administration of archive messaging folders. Without an SRM system in place to monitor, configure, trend, and automate the management of the files, a charge-back system would have been virtually impossible to develop or support. However, with SRM in place, charge back is possible, even practical.