Está en la página 1de 34

Best practices for defining

and managing your KPIs for


backup, recovery and
archiving
Session ID : BB2099
Romuald Boutault / Dec10, 2013
HP Storage Consulting Services
Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Context & Introduction


BURA
BackUp, Recovery and Archive

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Top IT spend priorities


BURA modernizations constitute 4 of the top 10 IT programs in 2013

Source : ESG Research Report, 2013 IT Spending Intentions Survey


4

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

BURA key challenges


Data protection and retention challenges

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

BURA key success factors


Classifying information is a pre-requisite for successful BURA projects

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

How HP BURA Consulting Services can help


Aligning Technology + Process + People to transform
Some backups
dont work

HP BURA
Consulting Services

Storage is too
Expensive

Integration
& Migration
Operational
Readiness

Restores are
not always
tested

Leverage consulting
to fix backups and
guaranty restores

PROTECTION
RISK
7

Strategy &
Architecture

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Leverage consulting
to enable archiving
solutions/services !

COST
COMPLIANCE

Not compliant
with legal
regulations

Generic approach to deliver consulting services


HP Consulting Services are based on a proven methodology based on a 5-step approach
Core Storage Consulting Services Activities (before solution implementation!)
Core Offering

Objective

Completion Criteria

1. Initiation (Kick-Off)

Validate the statement of work

Consensus obtained on
the statement of work.

2. Data Gathering (Discovery)

Collect and document the current environments


(architecture, pains, etc) and requirements!

Consensus obtained on current state including


prioritized pains and requirements!

3. Gap Analysis (Analysis)

Identify and analyze optimization opportunities

Consensus obtained on prioritized optimization


opportunities

4. Recommendations

Recommend vision, strategy, target architecture


and transformation path

Got buy-in on the recommendations conclusions

5. Business Case (Evaluations)

Evaluate the costs, the investments and ROI.

Got buy-in on the evaluations conclusions

Options that impact size/complexity/price of the service

HP workload?

<5 days

5 to 20 days

>20 days

Methods? Duration?

1-2 workshops

2-4-week Analysis

4-8-week Modernization

Scope?

Storage / Backup / Recovery / Archive

Drivers?

TCO reduction, Technology refresh, Data Classification

Discovery
tool?
NoThe
tool
or leverage
an internal/partner
eDiscovery
tool!
Copyright 2013
Hewlett-Packard Development Company, L.P.
information
contained
herein is subject to change
without notice.

Best practices
for defining and managing your
BURA KPIs
Samples of deliverables

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

What is the key success factor?


The key challenge is to 1) define, 2) report and 3) control the KPI
What types of KPI should be addressed ?
1) Technical KPI
2) Process-oriented KPI
3) Organizational-oriented KPI
What expertise/skills and maturity is required to manage KPI ?

10

1) Identify & Define (~Organizational)

Low

2) Measure & Report (~Tools)

Low

3) Control & Manage (~Process)

Low

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Medium
Medium
Medium

Target

What is your current and target maturity ?


Current

Key expertise (and associated KSF)

High
High
High

KPI#1 Technical indicators measured & reported


How do you measure and report your technical KPI ?
Are technical metrics defined? measured? monitored? reported?
# of clients (OS, active/inactive, Client Size/Throughput (Maxi, Average, Mini)
Backups (Window, Number, Duration, Capacity transferred)
Backups Success Rate (% of totally/partially completed in/outside windows, Failed with/without alerts)
Restores (Number tested/at request/after incident/disaster, capacity transferred)
Restores Success Rate (% of totally/partially completed according to RTO/RPO,
Type of media/drive utilization
Backup location (% of On-line, Externalized On-site, Offsite)
DB (Size, Max Utilization, Log Utilization, frequency of DB backups)
Drive Type (generation, breakdown)
Scheduling (Window, Number, Type (conditional or time), # of possible status)
Supportability of server and client software versions
Exclusions (Number of exclusions, % of total)
11

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#2 Process-oriented indicators documented


How do you document and implement your process-oriented KPI ?
Are Backup processes documented and operational ?
Provisioning Process
Change Management Process
Decommissioning Process
Legal Hold Process
Orphan Client Inventory Process
Testing Procedure

Are Restore processes documented and operational ?


Procedures for File Restores
Procedures for Full-System Restore
Procedures for Database Restore
Testing Procedure
12

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#3 Organizational indicators defined


How do you defined your organizational-oriented KPI ?
Standard roles and responsibilities related to data protection:
Role Definition for Backup Architect
Role Definition for Backup Administrator
Backup Responsibilities for Unix Administrator
Backup Responsibilities for Windows Administrator
Backup Responsibilities for Database Administrator
Responsibilities for Tape Handling / Operations

Are these roles and responsibilities :


Defined (drafted, completed, validated) ?
Documented (# of pages) ?
Measured and controlled (# of full/part-time FTE, fixed/controlled objectives/achievements) ?
13

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#4 Goals and objectives defined


How do you define your goals and objectives related to BURA services ?
Goals :
Find the best services/solutions/architectures to comply with your BURA service/operational levels
Understand business requirements related to BURA
Add Archiving in Service Catalog
Short-term objectives :
Understand best practices to build a BURA strategy
Define/test recovery scenarios (from simple to complex)
Define BURA service catalog based on best practices and IT maturity and get them
communicated/reviewed/validated by Business Lines
Mid-term objectives :
Design/Plan technology change while standardizing BURA processes/procedures to protect and archive
Long-term objectives :
Implement the new selected BURA services/solutions by replacing obsolete ones and creating new ones
while guaranteeing services levels
14

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#4 Goals and objectives defined


What is your current and target maturity level related to BURA services ?
This figure illustrates the steps involved in
maturing from a technology-centric organization
to an organization that harnesses technology as
part of its business strategy.

15

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Source: ITIL3 Service Strategy.pdf

KPI#5 Drivers and decision criteria set


What are your key drivers and decision criterias related to BURA services ?
The cost, the complexity of
the infrastructure and the
quality of the delivered
service are linked together

Control (reduce/
increase) the complexity :
Organization
Process
Technology
Low

Control (reduce/increase) the


costs :
Staffing
Equipments (HW, SW)
Environmental
Outages

Cost
H ig h

Low
H ig h

H ig h
I T E n v ir o n m e n t

Complexity
16

Control (reduce/ increase)


the quality:
Service levels
Client satisfaction

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Low

Quality

KPI#5 Drivers and decision criteria set


How do you define your driver and decision criteria related to BURA services ?
The reduction of the BURA TCO consists in reducing the optimal data size, storage
costs and inefficiency ratio

Data Archiving TCO = Size x Cost x Inefficiency Ratio


Cost [/GB/Year]

Reduce Storage Costs

(Storage TCO)

Data to retrieve
after a disaster
Reduce inefficiency
Size [GB]
(Optimal size of the data
to be archived)
17

Reduce size

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Inefficiency Ratio
(Extra capacity due to inefficiency)

KPI#5 Drivers and decision criteria set


How do you analyze the impact of your storage and BURA policies ?
IMPACT ANALYSIS OF STORAGE & BACKUP POLICIES

requires 11,6 GB of disk capacity

x11,6
x9,1

STORAGE

x7,8

X36
x11,6

x6,4
x2,7
x2,4

1 GB of production data

x1,4
x1,0

BACKUP

X36
x43

x12
x17

requires 43 GB (~100 usable-GB) of backup capacity


18

x2,4

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

x43

KPI#5 Drivers and decision criteria set


How to promote archiving services from a TCO standpoint !
Storage Services
TCO Analysis

In this example :
The Storage Services delivered 5 classes of services (Platinum to Bronze)
The average TCO vary from 65 to 1 /Used-Gb/year based on COS.

CURRENT

TARGET

Tier1
All data consolidated
a single storage
(mostly High-End)

Storage TCO
based on storage policy

Mini Storage TCO


(/Used-Gb/year)

Avg Storage TCO


(/Used-Gb/year)

Maxi Storage TCO


(/Used-Gb/year)
19

Tier2

Tier1
Most dynamic data stored
on original HE storage

Less dynamic data moved to


MR or LC storage

A2-Platinum2

A1-Platinum1

Tier3
Static Data migrated to archive services

B-Gold

C-Silver

D-Bronze

Storage
(Current)

A1-Platinum1

A2-Platinum2

B-Gold

C-Silver

D-Bronze

50

40

13

0,5

65

50

15

10

100

75

40

20

10

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#5 Drivers and decision criteria set


How to promote archiving services from a duplication standpoint !
Backup Services
TCO Analysis

In this example :
The Storage Services delivered 5 classes of services (Platinum to Bronze)
The data duplication ratio due to backup policy can vary from 350 to 3
TARGET

CURRENT
Tier1

Tier2

A1-Platinum1

A2-Platinum2

Tier1

Backup Policy

20

Tier3

B-Gold

C-Silver

D-Bronze

Storage

Storage

Class

Class

Class

Class

(Current)

(Target)

A-Platinum

B-Gold

C-Silver

D-Bronze

Partial Backup Frequency

Daily

Daily

Daily

Daily

Daily

Daily

- Retention

6 months

6 months

2 months

6 months

6 months

6 months

Full Backup Frequency

Daily/Weekly

Daily/weekly

1 month

1 annual

1 annual

1 annual

- Online retention

6 months

6 months

2 months

1 week

1 week

1 week

- Full (+offline) retention

7 years

12 months

12 months

1 year

1 year

1 year

# duplicated copies

200 < # < 350

48 < # < 170

12 < # < 48

12

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#6 BURA services distinguished and defined


How do you distinguish the different services related to Data Protection (BUR) ?
Your Data Protection services should be distinguished and defined :

21

To protect my data against

I manage an appropriate service :

Common/limited hardware failures

High-Availability / Business Continuity


(Device redundancy/fault tolerance with automatic failover)

Minor incidents and human errors

Operational Backups and Restores


(short/mid retention to restore files)

Major disasters

Disaster (Backups and) Recovery


(very short retention to restore an integrated system,
auditable and ready to be tested in a DRP exercise)

Time, Compliance, Regulations,

Archives and Retrievals


(long-term retention to retrieve an information)

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#6 BURA services distinguished and defined


How do you distinguish the different services related to Archiving ?
Data
Rationalization

Storage
Tiering

Automatic restore

Data
Archiving

Production

Archive1
Archive2

22

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Auto
Tiering

Low-Cost

Tapes

KPI#7 Your service catalogue summarized


Windows

Networker

Unix
V-LAN
Dual Path

Dual DC

Avamar

GOLD

SILVER

BRONZE

IRON

Mission Critical

Business Critical

Standard

Archives

BACKUP
Long-Term

SAN FC
Raid 5
Auto-Tiering
SSD-FC-SATA

SAN or NAS
D2D or Tapes
Static Tiering
SATA or Tapes

Y TB
SAN FC
Raid 5
Auto-Tiering
SSD-FC-SATA

Primary
backups

Primary
Copies

SAN FC
Raid 5
Auto-Tiering
SSD-FC

SAN FC or NAS
Raid 5 - 6
Auto-Tiering
FC-SATA

Passive

SAN FC
Raid 5
Auto-Tiering
SSD-FC

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

D2D Sata
Tapes LTO4

D2D Sata
Tapes LTO4

Local
Copies
(Clones)

Primary
Storage

SAN FC
Raid 5
Auto-Tiering
SSD-FC-SATA

Remote
Copies
(Passive)

BACKUP
Standard

Backup

Y TB

SAN FC
Raid 5
Auto-Tiering
SSD-FC

X TB

V-SAN
Single Path

LAN (CIFS, NFS)

Remote
Copies
(Clones)

V-SAN
Dual Path

Remote
replication
(Active)

Primary
Location
DR
location

Linux

HP-UX + Solaris

X TB

23

VMWare

Remote
replication

SAN

Hosts

How do you summarize and communicate your BURA service catalogue ?

D2D Sata
Tapes LTO4

D2D Sata
Tapes LTO4

KPI#7 Your service catalogue summarized


How do you summarize and communicate your BURA service catalogue ?
Hosts

Unix
Solaris 8-10 / Aix 6.1
Redundant SAN
Storage Policies

Policies

Windows
2003 / 2008

VMWare

STORAGE SERVICES

BACKUP SERVICES
Backup Policies

- 3 classes of storage proposed : 60% high-end, 30% Auto-Tiering & 20% mid-range
- Replication is optional
- Temporary local copies are proposed and optional (if unused storage capacity

- 1 or 2 cell manager(s)
- Oracle, Lotus, Sybase, FS, Copy
- More 150 backup tjobs/day

SAN Storage Production

SAN Backup

SAN Storage Test

Remote
copies

Remote
copies

Local
copies

Local
copies

Primary

Primary

SAN

Linux
RedHat

Performance
24

High, guaranteed

Dynamically Optimized

Normal, not guaranteed

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#8 Service and operational levels defined


How do you define your service/operational levels objectives/agreements ?
Service/Operational Levels
(SLA/SLO or OLO/OLA)

Tier1
Platinum

Tier2
Gold

Tier3
Silver

Tier4
Bronze

Hosts (Criticality)

Mission Critical

Business Critical

Mass Storage

Long-Term
Retention

Connectivity
(Attached, LAN, SAN, FCoIP, etc)

Edge-Core-Edge
Dual-Path SAN

Core
Dual-Path SAN

Core SAN
or FCoIP

LAN

Primary Storage
(Type, Performance, Availability)

Static
SSD+FC

Dynamic
SSD+FC+SATA

Dynamic
FC+SATA

Static
SATA

RAID10+3
Hourly Snapshots
3 days

RAID 6+2
Daily snapshots
1 week

RAID 50+1
n/a
n/a

RAID 5+1
n/a
n/a

High-Availability / Business Continuity


(type of redundancy, frequency, retention)
Operational Backups and Restores
(type, frequency, retention, RTO/RPO)

Daily (1 full / week - 5 incrementals / week)


Retention = 4 weeks
RPO~24h RTO~48h

DR Backups and Recovery


(type, frequency, retention, RTO/RPO)

Daily (1 full / week - 5 incrementals / week, retention = 3 days


RPO ~ 24h RTO ~ 1 week

Archives (type, frequency)

Hourly

Daily

Weekly

Weekly

Costs/Prices (/allocated-GB)

2,40

1,10

0,71

0,52

25

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#8 Service and operational levels measured

<10%
<50%
>50%

How do you report the distribution of your service levels ?


Service catalogue
SAN Connectivity

Storage

Operational
backups

Disaster Recovery

26

Platinum

Gold

Silver

Bronze

Accs

> 90%

Au moins 2 chemins daccs


indpendant au stockage avec dbit
garanti

0%

Au moins 2 chemins daccs


indpendant au stockage sans
dbit garanti

0%

Au moins 1 chemin daccs au


stockage sans dbit garanti

< 10%

Pas de chemin

Architecture

>90%

SAN volu

<5%

SAN Simple

0%

FC over IP

< 5%

NAS

Classe de
disques

<50%

> 4 Go/s

~25%

< 3 Go/s

~25%

< 1 Go/s

<1%

Aucune garantie

Disponibilit

~25%

> 99,99%

~50%

> 99,9%

0%

~25%

<1%

> 95%

Dlai
dallocation

>90%

Immdiat

0%

dans la journe

< 10%

Dans la semaine

>90%

Dans le mois

Dlai de rallocation

>70%

Dans la journe, si niveau de service


non respect

0%

A la demande et dans la semaine

<10%

A la demande et dans le mois

<20%

Non propose

Type de
sauvegarde et
restauration

<10%

A la demande et instantane avec


garantie de performance pour la
sauvegarde et la restauration.

<20%

A la demande et instantane
avec garantie de performance
pour la sauvegarde mais pas de
garantie pour la restauration.

>70%

Programme avec une


perturbation de linfrastructure
source minime et limite dans le
temps.

<1%

Programme avec
perturbation de
linfrastructure source
non maitrise.

Frquence

0%

1h

0%

12h

<90%

24h

<10%

1 semaine

Rtention

<5%

De 12 24 semaines

>90%

De 6 12 semaines

<5%

De 2 6 semaines

<1%

Moins d2 semaines

Type de
rplication

>90%

Rplication asynchrone sur longue


distance (>100 km)

0%

Rplication synchrone sur courte


distance (<100 km)

<10%

Rplication synchrone locale


(<10 km)

0%

Externalisation de bandes

PDMA/RPO

>90%

< 1 minute

0%

< 1 jour

0%

< 1 jour

<10%

= 1 jour

DIMA/RTO

0%

< jour

<90%

< 2 jours

0%

< 3 jours

<10%

> 3 jours

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#8 Service and operational levels defined


How do you define and communicate the RTO & RPO related to service levels
Objective : Decision matrix to decide what to recover after a disaster, and report the recovery status
Scope : Applications Service Level = GOLD (Mission Critical)
RPO
Data available to be recovered

RPO/RTO

~0

dd/mm/yyyy
hh:mm

~0

Host-based
mirroring

Checked

Storage-based
mirroring

Checked

DP solutions

24h

16h

12h

8h

4h

Local
snapshots

Checked

Remote
snapshots

Checked

Backups

27

RTO
Time to recover

Incident

To be
checked

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

20

1h

4h

12h

DRP Status
24h

>24h

Failed
Failed

Recovery
in progress

Ready

X
X

Check in
progress

KPI#8 Service and operational levels defined


How do you define the backup client profiles ?
Profile

Host Virtualization

SAN Storage

10

11

12

13

14

15

No

No

No

Small

Wintel

No Both

Yes

Yes

<4h

>4h

<4h

>4h

17

18

19

n/a

Unix

Infra (Standalone/Cluster)

16
Yes

Yes

Operating System

RTO/RPO (Y/N)

No

No

>4h

Solaris

Both

Yes

Yes

<4h

>4h

<4h

Aix

Wintel

n/a

>4h

<4h

>4h

<4h

n/a

Backup size (Small or Big Bck, PiTC)

28

Small Big

PiT

Small

Big

Small

Big

PiT

Small

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Big

Small

Big

PiT

Small

Big

PiT

KPI#9 Disaster Recovery Scenario documented

29

BRONZE

IRON

Mission Critical

Standard

Archives

RTO < 4h
RPO < 24h

RTO < 8h
RPO < 24h

RTO < 24h


RPO < 24h

BACKUP
Standard

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

Remote
Copies
(Clones)

Manually restore from


Networker backups via tapes

Remote
replication

RTO < 20
RPO < 1h

Local
Copies
(Clones)

Primary
backups

SILVER
Business Critical

Manually connect hosts


to replicated storage

Primary
Storage
Remote
Copies
(Passive)

GOLD

Remote
replication
(Active)

DR
location

Primary
Copies

Primary
Location

How do you document the data recovery scenario in case of disaster ?


BACKUP
Long-Term

KPI#9 Disaster Recovery Scenario documented


How do you document your data recovery procedures ?
A decent Data Recovery procedure should contain at least the following sections :
1.
2.
3.
4.
5.
6.
7.
8.
9.

30

Application Instance Overview


Data Protection and Recovery Strategy
Recovery Teams
Preparation Procedure
Shutdown Procedure
Failover Procedure
Startup Procedure
Acceptance Testing / Assurance Procedure
Handover Procedure

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#9 Disaster Recovery Scenario documented


How do you document your data recovery procedures ?

31

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

KPI#10 Data classification defined and analyzed


How do you discover the data types / profiles / classes ?
a) Total File Count (#) and Capacity (GB)

c) File Category Usage (pie


chart in %)

b) Predefined File Category with file


count (#), Capacity (GB) and Usage (%)
32

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

File type chart

KPI#10 Data classification defined and analyzed


How do you report the data types / profiles / classes ?
+48%
60% for
Email/Offices
Email
17,3TB

<0y => <3m

18% aged
> 1 year
33

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

vs 2010

Thank you

Copyright 2013 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.

También podría gustarte