is dcom more important than data centre design
TRANSCRIPT
1 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Is Data Centre Operations more important than Design ?
Presented by Barry Elliott of Capitoline
This is an extract from a webinar.
If you would like to see the full webinar please contact us at http://www.capitoline.org/ask-an-expert/
or would like a copy of the white paper that accompanies this presentation then please register at http://www.capitoline.org/is-data-centre-management-more-important-than-data-centre-design/
2 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
What is important to Data Centre Managers ?
• Reliability• Security• Efficiency• Maintenance• Capacity, growth and upgrades
3 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Reliability - Mean time to failure
From Capitoline’s recent research we found• 219 major data centre failures in 60 months, and that’s just the ones made
public• If we presume this is at best half of all failures then a data centre goes down
somewhere twice a week• And that’s excluding individual equipment failures• Average downtime 12.5 hours per major incident
• From 20 minutes to 8 days• Six of the data centres were considered written-off after fire or flood
4 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Failure mechanismsIf we remove IT problems we can look purely at the Facilities Management problems
Power43%
HVAC7%
Fire21%
Attack9%
Envi-ronment
6%
Un-known/Other
9%
Fire malfunction5%
5 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Design• Better design• More thought about location• Adequate fire suppression methodsOperations• Testing of all systems, not just components• Proper maintenance plans• Monitoring• Business processes
Almost every major failure can be avoided
6 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Designing for Reliability• TIA-942 describes four Ratings of fault tolerance• BICSI-002 and EN50600 describe four Classes of fault tolerance• The UpTime Institute describes four Tiers of fault tolerance
The engineering philosophy in each case is very similar
7 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Designing for resilience
The concept of Ratings/Classes and Tiers in a data centre
Rating 1. Enough items to do the jobRating 2. Redundant itemsRating 3. Concurrently maintainable.
• Any item or path can be taken down for maintenance and the system will still operate
Rating 4. Fault tolerant• The same as Rating 3 but will automatically cope with failures without human
intervention
8 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Managing for resilience
Not having an operational process plan means that even a data centre built to Rating 4 standards will not even operate at Rating 1 reliability levels as things will inevitably go wrong
Fatal Examples • A data centre failed because the standby generator was left in manual rather than automatic
mode. When the power failed, the generators didn’t start. It wasn’t anybody’s job to check this
• A data centre failed because the fire suppression system was set to manual rather than automatic. It burned down. It wasn’t anybody’s job to check this
• A data centre failed because temperature sensors were full of dust. Equipment went into thermal runaway. It wasn’t anybody’s job to check this
9 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Design and Management Standards
Design Standards
• EN 50600• TIA 942• BICSI 002
Management Standards
• ISO 9000• ISO 14000• ISO 27000• ISO 50000• ISO 55000
10 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
General ISO management standards aren’t enough
ISO 27000 isn’t enough on its own
ISO 9000Qualitymanagement
ISO 14000Environmental management
ISO 50000Energy management
ISO 55000Asset management
ISO 27000Information security management
DC Energy managementISO 30134EU Code of Conduct EN 50600-4
General management standards Data centre standards
You need EN50600 to prove compliance to Section 11 of ISO 27001/2 “Supporting utilities”
Note that ISO 45001 H&S, will be required. This is currently BS OHSAS 18001
11 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Main causes of failure
• Location• Building architecture• Computer room layout• Power • Cooling• Fire detection and suppression• Data cabling• Physical security• Building Management Systems
What’s important in data centre design?
Covered in detail by
EN 50600 and TIA942
No Single Points of Failure!
12 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
What’s important in data centre management?
• Understanding failure mechanisms• Good Housekeeping within the data centre• Optimised equipment layouts• Understanding the power train• BMS and status monitoring• Policies and procedures to manage a data centre• Essential maintenance• Health and safety requirements• Fire policy• Physical Security policy
13 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
They are both equally important. Without good design and good operations management we will not have good
• Reliability• Security• Efficiency• Maintenance• Growth and upgrades
Is Data Centre Operations more important than Design ?
14 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
physical plant maintenance arrangements
Day-to day operation of
the data centre
DR mode operations
Processes
Equipment & maintenance
manuals
Operations manual
Business continuity
and disaster recovery plan
Documentation
Building and operating a reliable data centre
Understand customer
requirements
Design the data centre
Build the data centre
Delivering the data centre
COMMISSIONING
Designer’s involvement
Operations
15 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
So how do you get the right balance ?
16 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
IndependentData Centre Consultants
We can help…
TrainingAuditDesignOperations
17 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
• Train yourself and your staff in datacentre design and operations
• Align your training with the International Standards
• Look for European Qualification Framework, EQF, approved training
But you can also help yourself…..
18 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Content:• Data Centre spaces• Raised access floors• Racks and location• Cooling• Power• IT grade Earthing• Cable containment• Fire systems• Cabling system design
3 day DCD - Data Centre Design
EN50600
Tiers & ClassesTIA942
EU Code of Conduct
PUEASHRAE
19 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Exercises:• Design a data centre building layout
• What rooms are needed?• Where should they go?
• Design the computer room layout• What is the most efficient rack layout?• Where should the cooling go to maximise efficiency?
• Calculate Cooling requirements• How much cooling do you need?
• Calculate the size of the power systems (generator, UPS etc.)• How much power do you need?• What size should the generator be?• What size should the UPS be?
3 day DCD - Data Centre Design
20 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
2 day DCOM - Data Centre Operational Management
• Understand what can go wrong• How you can prevent it happening• Learn how to improve
• Reliability• Efficiency• Security
• What procedures are needed?
21 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
2 day DCOM - Data Centre Operational Management
22 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Get Certified
+
=
23 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Capitoline’s DCE - Data Centre Expert qualification is approved as an European Qualification Framework Level 5 Qualification
Capitoline Data Centre Expert Qualification
BICSI recognise DCE training for Continuing Education Credits (CECs)
CIBSE recognise DCE training as Continuing Professional Development (CPD)
+ =
24 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
Matt FlowerdayBSC (Hons) Electrical and Electronic
EngineeringMBA
Over 30 years experience18 years owner/CEO of IT/Engineering
integratorFounding partners of Capitoline since 2006
Written training coursesTrained hundreds of data centre staff
Audited over 100 data centresMember of 50600 data centre standard
committeeDesigns and peer reviews of designs for data
centres in UK, Europe, Middle East and Africa
Barry Elliott Instructors
25 ©2015 Capitoline Ltd
DESIGN : OPERATIONS : AUDIT : TRAINING www.capitoline.org
This is an extract from a webinar.
If you would like to see the full webinar please contact us at http://www.capitoline.org/ask-an-expert/
or would like a copy of the white paper that accompanies this presentation then please register at http://www.capitoline.org/is-data-centre-management-more-important-than-data-centre-design/
Thank you for watching