bulletproof your linux environmentvox.veritas.com/legacyfs/online/veritasdata/sm b11.pdf ·...
TRANSCRIPT
Bulletproof Your Linux Environment
Chris Johnson
Senior Principal Technical Product Manager
Paul Belk
Director Technical Product Management
Bullet Proof Linux
Bulletproof Your Linux Environment
• Scale out workloads for faster performance
• Fast failover
• Reduce Storage Consumption
Future Proof Scalability
• Deploy HA/DR without complex and hard to maintain scripts
• Enable non-disruptive maintenance operations
• Robust support for server virtualization
Enterprise Grade HA/DR
Production
DR
• Improve operational scale
• Reduce storage costs with freedom of choice
• Migrate storage online with no application impact
Maximize Data Center Investment
2
Reduce Complexity Automate High Availability and Disaster Recovery
Metropolitan HA (Stretch Cluster)
Wide-Area DR (Global Cluster)
Local HA • Recover faster
• Reduce reliance on personnel during an
incident
• Reduce operator error
• Provide comprehensive data and application
availability
• Simplify by using a single solution for:
– Local HA
– Campus/Metro DR
– Global DR
The Benefits Symantec High Availability and Disaster Recovery
SAP APP 1 APP 3 APP 2 SAP SAP APP 4
Synch Replication or Mirroring
APP 4 APP 1 APP 2 APP 3
Asynchronous Replication
Bulletproof Your Linux Environment 3
From full upgrade to rolling upgrade Reduced application downtime during upgrade
Bulletproof Your Linux Environment
CFS
CVM
DMP LLT
GAB CFS
CVM
DMP LLT
GAB
HAD HAD
CFS*
CVM*
DMP* LLT*
GAB*
N1 N2
Old Version
New Version
CFS*
CVM*
DMP* LLT*
GAB*
HAD* HAD*
• Upgrade without any application downtime
• Two step process fully automated by Installer (CPI)
• Supported from 5.1 onwards for SFHA, CFS, CFSHA, SFRAC
Application Application
4
Simulate DR Failover: FireDrill
Test Your HA/DR Strategy With No Disruption
• Schedule and run actual application failovers at DR site
• Diagnose with built-in reporting
Primary Site
Volume Snapshot
Initiate Fire Drill Mount Snapshot
Replication
Test Application Reset
Secondary Site
Bulletproof Your Linux Environment 5
Accelerate Recovery: Intelligent Monitoring Framework (IMF)
• Registration, Blocking, Notification and Remediation
– Agent registers resource with the AMF Kernel Driver and provides it resource specific information
– Agent blocks and awaits notification on resource status change
– Resource status change instantaneously relayed to the blocking agent for handling
• Enable extremely fast failure detection and remediation
• Significant reduction in system resource utilization thus devoting key resources for other processing
HAD
Agent Framework
Intelligent Monitoring Framework (IMF)
Resources
IMF: Agents loosely coupled with resources, wait on
the AMF driver for notification
Registering…
AMF Driver
Faulted
Bulletproof Your Linux Environment 6
Maximize Availability Prevent Concurrencies
Node with Application Offline Node with Application Online
APP
Registering…
Application is online Application is Offline
VCS brings resource online
Agent registers with IMF
User attempts to bring app online outside of cluster
APP
IMF recognizes attempt
PCV prevents
• Works on Application Agent with VCS 6.0 • The only clustering technology with this feature • Preventing startup scripts is the future direction
• Prior to IMF-PCV, concurrency detection took up to 5 minutes
• No additional monitoring required as part of IMF framework
Bulletproof Your Linux Environment 7
Why Fencing?
Prevent Downtime Eliminating Corruption Risk with Fencing
APP APP
•Need to restrict writes to current and verified nodes
SCSI-3 Based Fencing CPS Based Fencing
SCSI3 PR Disks
• SCSI3 disks for i/o fencing
• Maximum data protection
• Non SCSI3 fencing
• Virtualized environment
Coordination Point Servers
Bulletproof Your Linux Environment
Data Corruption
APP APP
8
Prevent Downtime Control Which Node / Site Should Survive a Split-Brain
• Preferred fencing provides granular control to influence fencing outcome • Particularly useful for campus clusters and multi-node clusters where cluster
heartbeat goes over an actual network • Preference is configured as a ‘weight’ per node or service group
N1 (Active)
N2 (Standby)
N1 (Active)
N2 (Down)
Eliminate the Guess Work and Maximize Uptime
Bulletproof Your Linux Environment 9
Multi-Tier Applications & Services
Database
Application
Web server
Billing
app app app
DB
FS IP
VVR
Service Group
VCS
App
IP
Service Group
VCS
Web
VM
Service Group
Web
VM
Service Group
ApplicationHA
Veritas Cluster Server HA/DR
Veritas Cluster Server HA
Symantec ApplicationHA
Veritas Operations Manager
Bulletproof Your Linux Environment 10
Transition to Virtual Business Service Seamlessly
Database SG
Application Server SG
Web Server SG
Billing Virtual Business Service
Veritas Operations Manager
Start/Stop Order
Inter-cluster fault policy
Database
Application
Web server
Billing
app app app
DB
FS IP
VVR
Service Group
VCS
App
IP
Service Group
VCS
Web
VM
Service Group
Web
VM
Service Group
ApplicationHA
Bulletproof Your Linux Environment 11
Virtual Business Service Unparalleled Support Matrix
Operating Systems
Solaris
HP-UX
AIX
Linux
Windows
Virtualization
VMware
OracleVM, Zones
WPAR, LPAR
KVM
Hyper-V
High Availability
VCS 5.1, 6.0
VCS 5.1 SP1 (Unix)
VCS 5.1 SP2 (Win)
ApplicationHA 5.1 SP2
ApplicationHA 6.0
Databases
Oracle
Sybase
SAP
DB2
MySQL
Applications
Informatica
WebSphere
Microsoft Exchange
WebLogic
PeopleSoft
Replication
Veritas Replicator
Oracle Dataguard
EMC SRDF
NetApp SnapMirror
Bulletproof Your Linux Environment 12
Bulletproof Your Linux Environment
Robust Virtualization Support
13
VCS Guest Cluster for 24x7 Application Availability
24x7 Intelligent Monitoring by VCS
• Detailed application awareness
• Infrastructure/VM Availability
• Traditional clustering
Rapid Application Recovery
• Fast failover to standby VM
• Integrated with SFCFS
Minimize Application Downtime
• Protection against OS corruption
• No downtime for OS patching
Simplified Management
• App visibility through Veritas Operations Manager
• Fully supported with Virtual Business Services
VM1
OS
VM2
OS
HOST
SQL ORA
VM1
OS
VM2
OS
HOST
Web
VCS VCS
ORA
Bulletproof Your Linux Environment 14
ApplicationHA for application visibility and management
Simplify Application Management
• Manage app dependencies
• Start/Stop/Status of in-guest app
Fault detection and remediation
• Detailed application monitoring
• Application restart to remediate faults; no App/VM failover
Easy Configuration
• Automatic application discovery
• App wizards for configuring HA
• No private interconnects
Simplified Management
• In-guest app visibility through Veritas Operations Manager
• Fully integrated with Virtual Business Services for multi-tier application management
VM2
OS
HOST
ORA
ApplicationHA
VM1
OS
SQL
VM3
OS
APP
ApplicationHA
ApplicationHA
Bulletproof Your Linux Environment 15
Virtualization puts increased pressure on storage performance and availability
Bulletproof Your Linux Environment
• SAN aware load balancing key for high VM density
• Proactive failover of storage connectivity during disaster
• Flexibility to choose ANY cost effective storage
• Optimized handling of ANY tier of storage
• Visibility into storage characteristics needed
• Data center wide I/O bottlenecks critical to isolate
Performance & Resiliency
Storage Heterogeneity
Visibility & Reporting
RAID10, SRDF, Thin , SSD drive
I/O bottlenecks to storage arrays!
Lack visibility into performance bottlenecks!
Getting locked-in by single array vendor
Am I using the right tier of Storage? Device naming complicated in ESX!
16
Veritas Dynamic Multi-Pathing (DMP) for VMware: Improved storage I/O performance and availability
• Improve performance
– SAN & workload aware load balancing
– Real-time I/O performance graphs
• Increase availability
– Proactive path failover
– Automated path recovery
• Improve storage visibility
– Enhanced device naming
– Extended attribute discovery & reporting
– Data center storage reporting via vCenter
• Utilize a single multi-pathing solution across physical and virtual with heterogeneous storage
Bulletproof Your Linux Environment
Any X86 Server
VMware ESX
Hypervisor
ESX PSA Framework
MPP DMP
Guest Guest Guest Guest
17
VxDMP Components and Architecture
• vSphere / ESX Components
– Dynamic Multi-Pathing Core MPP components
– ASL / APM packages
• Management Interfaces
– vCenter integration for operations, visibility and reporting
– CLI Server for remote administration
• vCenter Functionality
– I/O policy management
– Path management (enable/disable)
– Performance graphs and tables
– DMP deployment statistics
Bulletproof Your Linux Environment
Storage
ESX Servers
vCenter Server
VxDMP plug-in in vSphere client
SAN VxDMP
Remote CLI
LAN
18
KVM: SFCFSHA in the KVM Host
Bulletproof Your Linux Environment
SFCFSHA in the Host:
•Replication of VM Images. •Highly available storage for virtual machines. •Simplified management of virtual machines. •Centralized Device Path Management
KVM Guest
RH6 KVM Host
KVM Guest
Shared Storage
Bo
ot
Dev
ice
App App
Dat
a D
evic
e
Bo
ot
Dev
ice
Dat
a D
evic
e
KVM Guest
KVM Guest
Bo
ot
Dev
ice
App App
Dat
a D
evic
e
Bo
ot
Dev
ice
Dat
a D
evic
e
SFCFSHA
RH6 KVM Host
19
KVM: SFHA in the Guest, DMP in the KVM Host
Bulletproof Your Linux Environment
Using SFHA in the Guest Provides:
•Enhanced database performance. •Point-in-time copy features for data back-up, recovery, and processing. •Consistent Device Naming Using DMP in the host provides: •Centralized multi-pathing functionality. •Active/passive array high performance failover. • Centralized storage path management. .
KVM Guest
RH6 KVM Host
KVM Guest
Shared Storage
Bo
ot
Dev
ice
App App
Dat
a D
evic
e
Bo
ot
Dev
ice
Dat
a D
evic
e
KVM Guest
KVM Guest
Bo
ot
Dev
ice
App App
Dat
a D
evic
e
Bo
ot
Dev
ice
Dat
a D
evic
e
RH6 KVM Host
DMP DMP
Storage Foundation + DMP
20
KVM Storage Management (SF 6.0)
Bulletproof Your Linux Environment
RHEL Host
RHEL Host
KVM KVM
RHEL Host
KVM
DMP
VM2
OS
SF
VM2
OS
SF
CVM/CFS
DMP
VM2
OS
SF
VM1
OS
SF
VM1
OS
DMP
• SF in the guest
– get consolidation benefits, keep
tools/processes unchanged
– enable P2V data migration with PDC
• DMP in the host
• CVM/CFS in the host
– present files (& filesnaps) as disks to guests
– enable live migration
– TP/R in the host
• Split stack (FS in guest)
• virtIO disk type support
• RHEL6
21
VCS High Availability in KVM Environment
In Guest Application Availability • Lean VCS footprint installed inside KVM Guest • No heartbeats, no additional network infrastructure • Start, Stop, Monitoring, Restart of Application • Support for Windows (6.0) and RHEL (5.1 SP1)
VM B
VM C
VM A
APP
KVM HOST
APP APP
VM Y
KVM HOST
VM X
APP APP
VM/HA without Application Awareness • VCS installed in KVM Host to form a cluster of KVM Hosts • Start, Stop, Monitoring, Failover of KVM Guest • No VCS footprint inside KVM Guest • Support for Windows (6.0) and RHEL (5.1 SP1)
VM B
VM C
KVM HOST
VM A
VM B
VM C
VM A
KVM HOST
VM/HA with Application Awareness • Lean VCS footprint in KVM Guest for Application Monitoring • VCS in KVM Host for KVM Guest & Infrastructure Monitoring • Application restart, VM failover (configurable levels) • Support for Windows (6.0) and RHEL (5.1 SP1)
VM B
VM C
KVM HOST
VM A
VM B
VM C
VM A
APP
KVM HOST
Bulletproof Your Linux Environment 22
Bullet Proof Linux
Bulletproof Your Linux Environment
• Scale out workloads for faster performance
• Fast failover
• Reduce Storage Consumption
Future Proof Scalability
• Deploy HA/DR without complex and hard to maintain scripts
• Enable non-disruptive maintenance operations
• Robust support for server virtualization
Enterprise Grade HA/DR
Production
DR
• Improve operational scale
• Reduce storage costs with freedom of choice
• Migrate storage online with no application impact
Maximize Data Center Investment
23
Bulletproof Your Linux Environment
Improve Operational Scale
24
Veritas Operations Manager (VOM)
Finance VBS
AppHA AppHA AppHA
Veritas Operations Manager
AppHA AppHA
Billing VBS
AppHA AppHA
HR VBS
VCS
VCS
VCS
CFS HA
SF HA
Finance Dept Billing Dept HRDept
Bulletproof Your Linux Environment 25
Improve Operational Scalability Unified management across platforms
Manage Veritas Cluster •Configure Service groups using wizards
•Perform availability operations using simple menus
•Monitor clusters and generate alerts
Manage Storage Foundation •Manage Disk Groups, Volumes and File Systems using wizards
•Monitor and setup copy services •Monitor and control volume replication
VOM
Bulletproof Your Linux Environment 26
Multi-Cluster Reports
Bulletproof Your Linux Environment
Cluster uptime analysis reports
One chart with service group
state across all clusters
Ability to click on the graph and do batch operations on service groups
27
VOM
Cross Cluster Reports
Bulletproof Your Linux Environment
Cross cluster failover summary
Date wise failover distribution
Service group wise failover distribution
28
VOM
Health Checks
Bulletproof Your Linux Environment
POLICY THEME Sample health check to include
1. Ensure availability frozen service groups
2. Maximize utilization underutilized disks and file systems
3. Enforce best practices Cluster OS consistency
Pick from 50 built-in health checks
Ability to write custom
health checks
Run & schedule health checks with a simple click
Recommend creating policies around the following 3 themes
29
VOM
Create custom templates for policies
Use Customizable signatures to comply with best practices,
availability and utilization policies
Bulletproof Your Linux Environment 30
VOM
Automate actions on faults
Configure rules to take action on fault
types
Bulletproof Your Linux Environment 31
VOM
Exploiting ANY Storage: Modular Architecture to Optimize to Any Storage
• Array Support Library (ASL)
– ASL Module for each Array (Ex: VMAX, NTAP)
– Discovers Array Mode Types (A/A, A/P, ALUA)
– Discover array specific attributes (Tier, RAID, Repl)
• Array Policy Module (APM)
– HW specific plug in to DMP
– Dynamically loadable module
– Understands Array specific failover behavior
– Allows Non Disruptive Upgrades for Controllers
• Allows DMP to support 50+ array families & 1000+ array models
Bulletproof Your Linux Environment
Any Storage
VMware ESX
Host Host Host Host
APM2
APM1
DMP
ASL1
ASL2
32
Improve operational efficiency and reduce complexity by using a single multi-pathing solution
Bulletproof Your Linux Environment
Dynamic Multi-Pathing Enables Storage Flexibility on all Platforms
• Leverage extensive array support
• Standardize on single multi-pathing across storage management stacks
• Seamless migration from native multi-pathing to Dynamic Multi-Pathing
• Choose storage based on price / performance needs, avoiding lock-in
Reduce operational costs using a single multi-pathing solution
Veritas Cluster Server
Volume Manger
File System
DMP
Veritas Cluster Server
DMP
Veritas Storage Foundation
ESX PSA Framework
MPP DMP for VMware
Veritas Application HA
Guest Guest Guest
VMware ESX
Guest
33
Bulletproof Your Linux Environment
Faster Migrations
Change with Minimum Disruption
34
Migration Steps Cross Platform Migration
Migration using traditional methods (network copy)
1. Provision Storage
2. Set up for copy process
3. Stop Oracle instance
4. Start copy process
– Using NFS, rcp/scp or similar
5. Wait for data transfer
– Can take many hours or days
6. Start Oracle instance
Bulletproof Your Linux Environment
Challenges with traditional methods: Requires double the storage
Require a high bandwidth network connection Will most likely require a long application outage to copy data
35
Migration Steps Cross Platform Migration
Migrating an Oracle Database between hardware platforms using Storage Foundation
1. Configure CDS
2. Stop Application
3. Validate file systems and convert byte order if necessary (to/from Linux)
4. Deport Disk Group
5. Import Disk Group on new host
6. Start Application
Bulletproof Your Linux Environment 36
Perform Storage Migrations Online Over any distance
Bulletproof Your Linux Environment
Migrate Arrays Locally Migrate Over Distance
• Wizard based configuration
• Configured during business Hours / Scheduled for non-Business hours
• Unattended Migration
• No Application Downtime
Veritas Storage Foundation SF, VCS
Synch/Asynch Replication
SF, VCS
• Performed Over IP without expensive Infrastructure
• Data Center Migration
• Allows for easy back out if issues detected
Utilize SmartMove to go Thin!
38
Bulletproof Your Linux Environment
Bulletproof Your Linux Environment
• Scale out workloads for faster performance
• Fast failover
• Reduce Storage Consumption
Future Proof Scalability
• Deploy HA/DR without complex and hard to maintain scripts
• Enable non-disruptive maintenance operations
• Robust support for server virtualization
Enterprise Grade HA/DR
Production
DR
• Improve operational scale
• Reduce storage costs with freedom of choice
• Migrate storage online with no application impact
Maximize Data Center Investment
39
CFS HA
Storage Foundation Cluster File System (SFCFS)
Bulletproof Your Linux Environment
CVM
High-performance Low-overhead architecture Scalable architecture Sub-minute failover
Highly-available Robust, long-lived product Supported by entire Storage
Foundation software stack Local, campus, or WAN cluster options
SAN storage
From 1 to 64 nodes
256 TB Max FS
40
SFCFS Use Cases: Scale-Out Applications
Bulletproof Your Linux Environment
Performance for Grid/Parallel Apps
• Minimal application overhead
• File access scales linearly as additional
nodes are added
• Single namespace for database copies
41
SFCFS
Maximize Availability / Minimize Complexity Eliminate Storage Failovers with CFS
Classic Clustering May Not Be Fast Enough Business Applications
Needing Maximum Uptime
Passive Server
Failed Server Active Server
Client
Active Server
Recovery Steps
• Detect failure • Un-mount file
system • Deport disk group • Import disk group • Mount file system • Start application • Clients reconnect
File System
Native clustering
File System
Native clustering
Bulletproof Your Linux Environment 42
Failover as Fast as Application Restart
Maximize Availability / Virtualize Confidently Eliminate Storage Failovers with CFS
Failed Server
Client
Active Server
Veritas Storage Foundation
Veritas Cluster Server
Veritas Storage Foundation
Veritas Cluster Server
Veritas Cluster File System
Business Applications
Needing Maximum Uptime
Recovery Steps
• Detect failure • Un-mount file
system • Deport disk group • Import disk group • Mount file system • Start application • Clients reconnect
Bulletproof Your Linux Environment
Hyper-V
43
Bulletproof Your Linux Environment
DB Clones
Compress De-Dupe
Compression Cloning & Snapshots De-duplication
Increase host storage efficiencies Compress where applicable Reduce storage consumption using
cloning and snapshot efficiently Reduce data duplication Enable thin provisioning & reclamation
Storage Foundation Cluster File System
Thin Reclamation
Store Information More Efficiently
44
Bulletproof Your Linux Environment
Bulletproof Your Linux Environment
• Scale out workloads for faster performance
• Fast failover
• Reduce Storage Consumption
Future Proof Scalability
• Deploy HA/DR without complex and hard to maintain scripts
• Enable non-disruptive maintenance operations
• Robust support for server virtualization
Enterprise Grade HA/DR
Production
DR
• Improve operational scale
• Reduce storage costs with freedom of choice
• Migrate storage online with no application impact
Maximize Data Center Investment
45