Protect Your Database: High Availability for High Demand Data

45
Grab some coee and enjoy the pre-show banter before the top of the hour!

Transcript of Protect Your Database: High Availability for High Demand Data

Page 1: Protect Your Database: High Availability for High Demand Data

Grab some

coffee and enjoy the pre-show banter

before the top of the

hour!

Page 2: Protect Your Database: High Availability for High Demand Data

H T Technologiesof 2016

Page 3: Protect Your Database: High Availability for High Demand Data

HOST:EricKavanagh@eric_kavanagh

Page 4: Protect Your Database: High Availability for High Demand Data

THISYEARis…

Page 5: Protect Your Database: High Availability for High Demand Data

THELINEUP

ANALYST:

DezBlanchfieldDataScientist,TheBloorGroup

ANALYST:

RobinBloorChiefAnalyst,TheBloorGroup

GUEST:

BertScalzoSeniorProductManager,IDERA

Page 6: Protect Your Database: High Availability for High Demand Data

INTRODUCING

RobinBloor

Page 7: Protect Your Database: High Availability for High Demand Data

Robin Bloor, PhD

Page 8: Protect Your Database: High Availability for High Demand Data

The Service Level Dynamic

Users NEVER ask for lower service levels

Page 9: Protect Your Database: High Availability for High Demand Data

Service Level Definitions

u  Service levels usually defined in

terms of application performance,

availability and metrication

u  They move, often (unfortunately)

because they are compared to

Internet services

u  Availability for databases includes

BU & Recovery, and DR

u  Business processes have service

levels and they depend upon IT

service levels being met

Page 10: Protect Your Database: High Availability for High Demand Data

Acceleration & Time Barriers

u  We are gradually moving to

event/real-time processing and

thus all applications are

gradually accelerating

u  The time barriers are:

•  AFAP

•  0.1 seconds

•  4 sec to hours

•  Day/Week/Month

u  Technology change is a factor in

this

Page 11: Protect Your Database: High Availability for High Demand Data

Layers of Complexity

BusinessServiceLevels

ApplicationServiceLevels

InfrastructureServiceLevels

Non-Stop HotStandby

WarmStandby Reboot

Configurations

ServiceLevelD

ependencies

Page 12: Protect Your Database: High Availability for High Demand Data

The Complexity Curve

u  Complexity derives from

nodes and connections

(dependencies) – which are

(likely to be) growing

u  For availability levels there

are time targets – which are

(likely to be) reducing

u  The natural evolution is

therefore towards non-stop

operation

u  Non-stop is expensive

Page 13: Protect Your Database: High Availability for High Demand Data

The Net Net

The management of service levels is

an on-going activity

Page 14: Protect Your Database: High Availability for High Demand Data

INTRODUCING

DezBlanchfield

Page 15: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

HighAvailabilityDatabaseChallenges

Page 16: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Whatdowemeanwhenwe

talkaboutdatabasesystems&

databasepla;ormavailability

Page 17: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

WhatdoesitmeantothinkaboutHighAvailability,isthereasilverbulletorisitalotmorecomplexthatthat..

§  Isitup>me(i.e.five9’s)

§  Whataboutservicelevelagreements

§  Loadbalancing&Faulttolerance

§  Ac>ve/StandbyorAc>ve/Ac>ve

§  Mul>pleservers&storage

§  Redundantreplicatednetworks&datacenters

§  ArewetalkingaboutN+1

§  Nosinglepointoffailure

WhatdowemeanbyHighAvailability

!?

Page 18: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Whenwetalkaboutdatabaseavailabilityweusuallyenduptalkingaboutpercentagesinnines..

§  OneNine=(90%)=36.5daysdown>mep/yr

§  Twonines(99%)=3.65daysdown>mep/yr

§  Threenines(99.9%)=8.76hoursdown>mep/yr

§  Fournines(99.99%)=52.56minsdown>mep/yr

§  Fivenines(99.999%)=5.26minsdown>mep/yr

§  Sixnines(99.9999%)=31.5secondsdown>mep/yr

Whatdifferencecanasingleninemake

%

Page 19: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Thereisnosinglesilverbullet

whenitcomestoavailability,

andonesizedoesnotfitall

Page 20: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Availability§  Ameasureofthe>methataserverorprocessfunc>onsnormallyforgeneral

usage,aswellasameasureoftheamountof>methattherecoveryprocess

requiresaberacomponentfailure.

HighAvailability

§  Systemdesignandimplementa>onthatachievessystemanddataavailability

almostallofthe>me,24hoursaday,7daysaweek,and365daysayear.High

availabilitydoesnotequateto100%availability.Toachieve100%availabilityis

notacost-effec>verealityforthelargemajorityofimplementa>onstoday;

rather,itisagoal.

AvailabilityInAPlainEnglishDefini>on

Page 21: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Cost&ComplexityVSTypeofservicebeingdelivered..

AvailabilityVSTypesofService

Page 22: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

ThetradeoffbalancebetweenHigh&ConDnuousavailability..

InvestmentVSReturninAvailability

Page 23: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

MeanTimeBetweenFailure

Or

MeanTimeToRecover

Page 24: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

WhenitcomestodatabaseandsupporDnginfrastructureHighAvailability,everyonehastheirownneedsandwants..§  Whatcanyourorganisa>onafford

§  Whatcanyourbusinesssupport

§  Whatskills&technologydoyouhavein-house

§  Canyououtsourcesomeofthechallenge

§  Whattypeofmarketorcustomerareyouservicing

§  Mee>ngproduct/service/industrynorms

§  On-prem/externallyhosted/inthecloud

MyAvailabilityisnotYourAvailability

!!

Page 25: Protect Your Database: High Availability for High Demand Data

@dez_blanchfield

Matchingdesiredorrequired

availabilitywithperformance&

businessneedsisnomeanfeat,it

takes>meeffort&money

Page 26: Protect Your Database: High Availability for High Demand Data

INTRODUCING

BertScalzo

Page 27: Protect Your Database: High Availability for High Demand Data

© 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. © 2016 IDERA, Inc. All rights reserved.

PROTECT YOUR DATABASE: HIGH AVAILABILITY FOR HIGH DEMAND DATA

Bert Scalzo Senior Product Manager

Page 28: Protect Your Database: High Availability for High Demand Data

2 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 2 © 2016 IDERA, Inc. All rights reserved.

AGENDA

§  Goal: Database Agnostic

§  Overall Business & Technology Perspective

§  Business Requirements

§  Most Basic Pseudo High Availability Solution

§  Storage Level Solutions

§  Virtualization Level Solutions

§  Database Level Solutions

§  Selection Criteria

Page 29: Protect Your Database: High Availability for High Demand Data

3 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 3 © 2016 IDERA, Inc. All rights reserved.

GOAL: DATABASE AGNOSTIC

§  All major RDBMS vendors offer something §  There are some 3rd party software vendors §  No one solution is best – it all depends… §  One universal fact – more moving parts, thus

more complex and possibly costly

§  Note – all examples generic, not DB specific

Page 30: Protect Your Database: High Availability for High Demand Data

4 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 4 © 2016 IDERA, Inc. All rights reserved.

OVERALL BUSINESS & TECHNOLOGY PERSPECTIVE

§  Data is the most important asset

§  Fast access to data is always nice

§  Reliable access to data is critical (need for high availability or HA)

§  Data downtime can cost a fortune – lost revenue & lost customers

§  Since hardware now so cheap, more & more demand for high availability

§  Offerings from storage vendors, database vendors and virtualization vendors

§  With emergence of the cloud, high availability can be as simple as a checkbox

Page 31: Protect Your Database: High Availability for High Demand Data

5 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 5 © 2016 IDERA, Inc. All rights reserved.

WHAT ARE THE BUSINESS REQUIREMENTS?

§  What is the scope of failure protection? •  Within a single site •  Across multiple sites

§  What is the level of failure protection? •  Power failure •  Component failure •  NIC/HBA •  Hard Disk

•  Storage failure •  Computer failure •  Site failure

§  What is the expectation for time to resume operations?

§  What is the expectation for data loss upon resumption?

Page 32: Protect Your Database: High Availability for High Demand Data

6 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 6 © 2016 IDERA, Inc. All rights reserved.

MOST BASIC PSEUDO HA SOLUTION

§  Not a complete HA solution §  Common minimalist architecture due

to budget or resource constraints §  No special skills or admin costs §  Only protects single site against: •  Power Failure •  UPS

•  Component Failure •  Bonded NICs •  Multipath HBAs •  Dual controllers •  RAID-10 disks

Page 33: Protect Your Database: High Availability for High Demand Data

7 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 7 © 2016 IDERA, Inc. All rights reserved.

STORAGE LEVEL SOLUTIONS

§  Not a complete HA solution §  Primarily the responsibility of

the storage admin §  There may be multiple

options – see common examples to the right à

§  RAID-1 across SANs may not be universally available

§  Storage virtualization can come from multiple sources

§  Storage replication means recovery at the block or disk level, and not at the database or transaction

B ond in g

R A ID -0 R A ID -0

R A ID -1

Storage virtualization Storage replication

Page 34: Protect Your Database: High Availability for High Demand Data

8 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 8 © 2016 IDERA, Inc. All rights reserved.

SERVER VIRTUALIZATION LEVEL SOLUTION

§  Not a complete HA solution §  Primarily the responsibility of the

virtualization admin §  There may be multiple options (key one

to the left) §  Some vendors offer a live relocation of

some database while others do not

Page 35: Protect Your Database: High Availability for High Demand Data

9 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 9 © 2016 IDERA, Inc. All rights reserved.

COMBINING STORAGE & SERVER VIRTUALIZATION

§  Finally a complete HA solution §  Primarily the responsibility of the storage &

virtualization admins §  Some vendors offer a live relocation of

some database while others do not §  Multiple vendors for a single issue §  More moving parts that are not necessarily

in sync as a whole §  But virtualization at all levels is very, very

popular as a general solution…

Page 36: Protect Your Database: High Availability for High Demand Data

10 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 10 © 2016 IDERA, Inc. All rights reserved.

SHARED STORAGE SCALABLE CLUSTER

§  Not a complete HA solution §  Primarily the responsibility of DBAs §  Not all database vendors offer this option §  Oracle Real Application Cluster (RAC) §  Costs additional money to both buy and

administer (need higher skill set) §  Can address two issues: •  Scalability •  High Availability

B ond in g

Mu lt i -P a th

R A ID -1 0

B ond in g

Page 37: Protect Your Database: High Availability for High Demand Data

11 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 11 © 2016 IDERA, Inc. All rights reserved.

SHARED STORAGE FAILOVER CLUSTER

§  Not a complete HA solution §  Primarily the responsibility of DBAs §  Varies by name and features across

different database vendors §  But essentially this is it…

Page 38: Protect Your Database: High Availability for High Demand Data

12 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 12 © 2016 IDERA, Inc. All rights reserved.

REPLICATION - PHYSICAL

§  Finally a complete HA solution §  Primarily the responsibility of DBAs §  Varies by name and features across

different database vendors §  But essentially this is it…

Page 39: Protect Your Database: High Availability for High Demand Data

13 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 13 © 2016 IDERA, Inc. All rights reserved.

REPLICATION - LOGICAL

§  Finally a complete HA solution §  Primarily the responsibility of DBAs §  Varies by name and features across

different database vendors §  But essentially this is it…

Page 40: Protect Your Database: High Availability for High Demand Data

14 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 14 © 2016 IDERA, Inc. All rights reserved.

COMBINING DATABASE & STORAGE VIRTUALIZATION

§  Finally a complete HA solution §  Primarily the responsibility of DBAs

and storage admins §  Multiple vendors for a single issue §  Especially good for major company

with different geographic locations

§  Note – this example is quite easy to set up in the cloud since it’s just a few checkboxes and selections to make this picture work

Page 41: Protect Your Database: High Availability for High Demand Data

15 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 15 © 2016 IDERA, Inc. All rights reserved.

THESE ARE JUST THE BASICS

§  There are more combinations than shown here

§  New technologies and options emerging regularly

§  High Availability far more common than just 10 years ago

Page 42: Protect Your Database: High Availability for High Demand Data

16 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 16 © 2016 IDERA, Inc. All rights reserved.

SELECTION CRITERIA

§  No golden rule §  It all depends §  Ask business expectations §  See these examples §  Pick what fits best §  Take your time to set up §  Monitor to know what’s

correct

Page 43: Protect Your Database: High Availability for High Demand Data

17 © 2016 IDERA, Inc. All rights reserved. Proprietary and confidential. 17 © 2016 IDERA, Inc. All rights reserved.

THANKS! Any questions?

You can find me at: [email protected]

Page 44: Protect Your Database: High Availability for High Demand Data
Page 45: Protect Your Database: High Availability for High Demand Data

TheArchiveTrifecta:•  InsideAnalysiswww.insideanalysis.com•  SlideSharewww.slideshare.net/InsideAnalysis•  YouTubewww.youtube.com/user/BloorGroup

THANKYOU!