Building Enterprise Class Hyper-V Solutions with Symantec and Microsoft

(1)

Building Enterprise Class Hyper-V

Solutions with Symantec and Microsoft

(2)

(3)

Competitive Landscape

Hypervisor

War

Native

Technology

Incumbents

& Startups

Growing competition for VMware in the virtualization

market space

Microsoft continues Hyper-V growth, catching

up to VMware

Virtualization Wars: Vmware vs. Hyper-V:

Which is Right For Your Virtual Environment?

SMB 3.0 brings enterprise-class storage to

Hyper-V deployments

Windows Storage Spaces and ReFS: Is it time

to ditch RAID for good?

EMC XtremeSF: Delivering Next Generation

Storage Performance for SQL Server

Running SQL Server on all-flash crushes storage latency,

enabling 3-5x better transactional thoughput and

(4)

So is it really

Almost, so what are the real challenges?

(5)

• Increasing VM (& app) consolidation ratio

• Storage growth, Increasing LUN sizes

• Create, delete, migrate of VMs & storage

• Maximize Application Availability

Challenges in Windows Hyper-V Environments

VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM

App App _App _App _App _App _App _App _App _App _App

VM VM VM VM VM VM VM VM VM

IO Bottlenecks

App

lic

at

ion

U

pt

ime

VM Cost-effective VM/application centric storage management

Storage Pooling

Leverage Flash and improve latency and throughput in physical/virtual

I/O Performance

Maximize Application uptime

Application uptime

(6)

(7)

• Dramatically lowering the costs and

effort of delivering IaaS storage

services

• Disaggregated compute and storage

• Independent manage and scale

at each layer

• Industry standard servers,

networking and storage

• Inexpensive networks

• Inexpensive shared

JBOD storage

Infrastructure-as-a-Service Storage Vision

2x Increased throughput

3x Reduced cost

5x Increased capacity

6x Reduction in servers

Po w erSh ell & SCVMM 2012 R2 Ma n agem en t Hyper-V Clusters

Scale-Out File Server Clusters

Storage Spaces Virtualization and Resiliency SMB

(8)

with FC/iSCSI Storage Array

• Storage Tiering

• Data deduplication

• RAID resiliency groups

• Pooling of disks

• High availability

• Persistent write-back cache

• Copy offload

• Snapshots

Comparison with Hardware Array “Enterprise-Grade”

Capabilities

with File Server & Storage Spaces

• Data deduplication

(enhanced with

R2)

• Flexible resiliency options

(enhanced

with R2)

• Pooling of disks

• Continuous availability

• Persistent write-back cache

(9)

Storage Pools Capabilities

Microsoft

• Pooling of disks

• Flexible resilient storage spaces

• Native data striping

• Enclosure awareness with certified

hardware

• Parallelized rebuild utilizes spare pool

capacity

• Data Integrity Scanner

• Periodic background scan detects

latent corruptions, auto-corrects

where possible

• Integration with both NTFS and ReFS

• Thin Provisioning

• Deduplication

Physical Disks from Shared SAS JBODs

Storage Pools

Storage Spaces

(10)

Storage Foundation Windows v/s Microsoft Storage Pools

Storage Protocol Support

• Supports only SAS based storage.

• Does not support FC or iSCSI

http://technet.microsoft.com/en-us/library/jj822938.aspx

Install & Upgrade

• Disks added to Storage Pools get formatted.

Existing data gets lost

• Costly and complex way to preserve existing

data –

1. New disks added to a storage pool.

2. Xcopy data from the original disks

Enclosure Awareness

• Complex and limited protection from array

failures.

Storage Migration

• Depend on ODX support from array vendors to

do online storage migration.

Storage Protocol Support

• Supports FC, iSCSI, SAS, SSDs

• Use the existing SAN or new DAS infrastructure

with equal easy

Install & Upgrade

• Application data stored in a SFW Dynamic disk is

forward compatible. Data stored with SFW

5.x/6.x can be directly accessed after upgrade.

Enclosure Awareness

• Application data can be mirrored across arrays

to protect from enclosure failures.

• Grow/shrink of volumes are done respecting the

array boundaries.

Storage Migration

• Migrate application data from one enclosure

online – no vendor specific support required.

Microsoft Storage Pools

Storage Foundation for Windows

(11)

Storage Management for Hyper-V Virtual Machines

Storage Pools on Windows

• Storage Pools created under the FailOver Cluster • Volume as a unit-of-failover

• VHDs stored on resilient volume

Seamless integration into Ecosystem

• Failover Cluster aware storage pooling

• Integrated with native management workflows (FoC Mgr, SCVMM, Hyper-v Mgr, Server Mgr)

• Supporting all native Hyper-V VM capabilities ( e.g. Live/Quick Migration)

• Live Migrate Hyper-V Virtual Machine

• VHDs would lie on SFW Volumes

• Coexist with CSV

• 1 VM mapped to SFW Volume(s).

Advanced Storage Capabilities

• Protect from Storage hardware failures • Vendor independent Live Storage Migration • Capacity Monitoring

• Site aware storage management

Architecture

Storage Foundation Storage Pools

Symantec

SFW SFW SFW

Vmdell-12 Vmdell-13 Vmdell-15

FoC Cluster

SFW Storage Pools

Data VHDs

Cluster Shared Volume

Boot VHDs

(12)

How SFW helped Hyper-V win over VMware

• Provide a low cost scale-out virtualization solution • Provide DR with least RPO

• Protect against array failures

VMware based Solution

Business Objective

Costly

Hypervisor : VMWare

Storage : EMC Array (VMAX)

Replication : SRDF license

DR Orchestration : Site Recovery Manager (SRM)

Or

vMSC Certified : VPLEX/SVC/NetApp/Dell

Drawbacks with Competing Solution

• Create a Stretched Cluster (Campus Cluster) with FoC and SFW

• No separate Replication product required • No separate Orchestration required • No “certified solution” required • Hyper-V less costly than VMWare

Winning Points with SFW + Hyper-V

(13)

(14)

SFW SmartIO v/s Tiered Storage & Write-Back-Caching

IO Performance

• ( Work in Progress)

• ( Most of the numbers shown by Microsoft are

not OLTP type workloads)

Failure Protection

• In the PCIe SSD fails – there is application data

lost.

• If the SSD is mirrored – the IO performance is not

predictable.

• In case of a node failure, if the data resides on

PCIe SSD, that data will not be available on the

other nodes.

• Write Back Caching is not cluster aware

Live Migration Support

• Can Storage Tiering & Write-Back-Caching

support VM Live Migration?

IO Performance

• Physical : 3x improvement in IO performance

under a true OLTP workload - TPCe

• Virtual : 2.5x improvement in IO performance

when SQL hosted inside Hyper-V VMs (each

running TPCe)

Failure Protection

• If the PCIe SSD fails, application data is not lost

as the data is written to the HDD.

• In case of Clustered environment, a node loss

would not result into application data lost

Live Migration Support

• LM will continue to work as VM data is always

available on all nodes of the cluster

Tiered Storage & WBC

SmartIO

(15)

SmartIO: SSD as a Cache

Block Level Read Cache

• Intelligently cache reads.

• Write-Through Cache – no data-loss in case of SSD failures.

• Will work in a clustered environment

Deployment

• Enabled for all volumes

• No application reconfiguration required – transparent & seamless integration

Operations

• Online Enable/Disable cache • One Default cache for all volumes

(16)

SmartIO: SQL Consolidation

Address IO logjam due to SQL or VM consolidation

• Multiple databases in a single SQL instance.

• Multiple SQL instances on a single physical

server

• Single instance SQL hosted inside Hyper-V

Virtual Machines

Improve SQL IO Performance in a clustered

Environment

• Internal (PCIe) SSD drives used as storage

cannot be used as shared storage

(17)

TPCe Configuration :

• 4VM with SQL

• SSD size: 1.2 TB

• Database of 160k

customer.

• OLTP workload of

50 users (per VM).

TPCe Configuration :

• 4 SQL instances on

single server.

• SSD size 1.2 TB.

• Database of 160k customer

• OLTP workload of 50 users

(per SQL instance)

SmartIO :Cost – Performance Analysis

(18)

Auto discovery

• FusionIO & Intel in-built support for autodiscovery

• CLI to tag a device as SSD

Single instance SQL database

• Log files on the SSD

• Lot of sequential writes and reads – better on

SSD

• Indexes on SSD

• Lot of random reads – better on SSD

Protecting from Failures

• Mirror it with HDD. Make it the ‘preferred read

plex’

• HDD plex could be a Shared disk in case of cluster

DR

• Single node VCS cluster (supports non-SCSI devices)

(19)

Storage Management for Hyper-V Virtual Machines

Cost Effective & Resilient Storage Pooling for Hyper-V

• Failover Cluster aware

storage pooling

• Integrated with native

management workflows

(FoC Mgr, SCVMM, Hyper-v

Mgr, Server Mgr)

• Supporting all native

Hyper-V VM capabilities (

e.g. Live/Quick Migration)

• Protect from Storage

hardware failures

• Vendor independent Live

Storage Migration

• Capacity Monitoring

• Site aware storage

management

• Wizard driven storage

provisioning &

management

• VHDs on resilient storage

volumes.

• Online grow and shrink of

volumes

• VSS aware snapshots

Seamless Integration with

Ecosystem

Advanced Storage

Capabilities

Simplified Storage Services

(20)

(21)

(22)

VM Monitoring in Windows Server 2012

Overview

• Application Health detection inside virtual machine

• Cluster service in host takes remedial action

• Independent of Guest Clustering

• No need for clustering in guest

• Windows Server 2012 Required

• As both host and guest OS

• Windows Server 2012 Hyper-V integration services on guest

(23)

Coordinated recovery with Hyper-V Role and Microsoft

Failover Cluster – Details

• Coordinated handoff to Failover

Cluster Resource Policies

–

VM level recovery and failover for

resource faults.

–

Granular level settings per VM

• Utilizing Microsoft Heartbeat

service within Failover Cluster

–

Leverage heartbeat service for

communication of faulted applications.

–

Can be enabled and disabled from

(24)

Providing application resilience with Microsoft FOC and

Symantec ApplicationHA

Application

Resources

OS

VM

Hyper-V

Server

Site

App

Components

VM1 OS VM2 OS

Microsoft Failover Cluster

(25)

VM1 OS VM2 OS VM1 OS VM2 OS

Coordinated recovery between ApplicationHA and

Microsoft Failover Cluster

• App monitoring

–

Show health status

–

Detect app failures

• Coordinated recovery

–

Restart applications

–

Trigger Failover Cluster

for further recovery

–

Integration via VM

Heartbeat Service

• Protects against wide

range of failures

–

Infrastructure failures

–

VM is up but app is down

Microsoft Hyper-V Microsoft Hyper-V Application

HA Application HA IIS

SQL SQL IIS

Failover Cluster Failover Cluster

–

VM recovers after a server failure but app doesn’t

–

App is up but not functional

VM1

OS

SQL Application

(26)

OS

Coordinated recovery between ApplicationHA and

Failover Cluster – Details

• Deep understanding of the apps

–

App specific modules start, stop, recover

apps

–

Eg: SQL DB instances, FileStream,

Analysis, Storage mount points,

dependencies

–

Functional testing based on SQL queries

• Customizable recovery behavior

–

App or VM restart limit

–

In-guest remediation only

-

Enable/Disable App Heartbeat

–

Turn off remediation during planned

maintenance

-

Enter/Exit Maintenance mode

(27)

Platform and Application Coverage

ApplicationHA Guest

• Platform Support includes

• Windows Server 2008 R2, 2012, 2012 R2 (all x64)

• Wizard driven auto discovery of applications include

• Custom Application

• SQL Server 2008

• SQL Server 2012

• Exchange Server 2010

• IIS 6.0, 7.0, 8.0, 8.5

• Sharepoint 2010

• Oracle Database 11gR2, 12c

• FileShare

• PrintShare

• Can also leverage Symantec Cluster Server agents

(28)

Coordinated recovery between ApplicationHA and

Microsoft Failover Cluster – Details

• Simple Application Configuration Wizards

(29)

Visualize and control apps from web browser -

Screenshot

Detailed Monitoring Application Operation

(30)

• Coordination between application tiers drives the business

service for its organizations and for its customers

• Reduce operational headaches

• Orchestrated start/stop order.

• Auto Fault remediation to reduce siloed management headaches.

Applications Working Together Drives Your Business!

Message Service

Web Service Database

Audit Log

Business Application

2

1

3

(31)

SYMANTEC VISION 2014

Should it take hours to recover from a failed service?

Virtual Business Services Lab

00:00

Database faces an outage, automatically recovered on standby node

00:01

Service fails, customers lose access to balance enquiry website

00:05

Monitoring software detects outage, creates ticket

00:10

Help desk starts processing ticket, finds service owner

00:20

Service owner starts troubleshooting: are web servers OK?

00:30

Locate the virtualization admin: troubleshoot VM layer

00:45

Virtualization admin: Web Servers OK, not my problem, check DB

00:50

Look up dependencies, service is using CustomerInfo database

01:00

Pull up the DBA: Is the database working OK?

01:15

DBA: DB all clear, my databases are all well protected!

01:30

Continue troubleshooting: DB OK, Web OK, is BusinessLogic App OK?

01:40

BusinessLogic Application not touched in ages, what’s the password?

02:00

Oh no, the Application is pointing to the failed DB server …

02:10

Reset the Application to point to the active DB server

02:15

Check if website is accessible … Yes? You’re sure? PHEW!

(32)

Virtual Business Service

Single logical entity for the multi-tier application

Veritas Operations Manager

Database

Application Server

Web Server

Virtual Business Service

(33)

Always-on automatic multi-tier protection with Virtual

Business Services

Database

Application

Web server

app app app

Business Application

VCS ApplicationHA

Automatic application

protection

Veritas Cluster Server

Automatic fault & recovery

communication

Virtual Business Services

Configurable reaction to

lower tier events

Virtual Business Services

00:00

Database faces an outage

00:00

Website down

00:01

VCS recovers Database; DB OK

00:03

VBS transmits DB recovery to App

00:04

Application reconnects; App OK

00:05

Website Up!

(34)

Virtual App Tier

Hyper-V Failover Cluster

Utilize current infrastructure to provide

service level orchestration

Ap p licatio n H A Ap p licatio n H A Database (MS FOC) Standalone Application Server Web Servers

Virtual Business Services

Introducing Virtual Business Services for Microsoft

Failover Cluster

Proactive Service health validation –

lower operational expenses

Overcome operational silos, eliminate

availability issues quicker

(35)

Maximize Application Uptime

Making Hyper-V Enterprise Ready

• Single Multipathing

solution for multiple

array vendors

• Microsoft WHQL

certified

• Quarterly updated

support for new arrays.

• Campus-level DR for

Virtual Machines with Zero

RPO

–

Store Virtual Machine data

mirrored volumes that are

site –separated.

• No extra license for a

replication software

• No replication

management.

• Monitor Application

and remediation hosted

inside a Hyper-V VM

• Visibility and managing

application

dependencies

• Support for various

applications inbox

Dynamic Multipathing

Zero RPO DR solution

Application HA

Cost Effective & Resilient VM specific storage services

VM 1

(36)

(37)

Building Enterprise Class Hyper-V Solutions with Symantec

Storage Foundation High Availability for Windows 6.1 :

GA

Ensure 24x7 Availability

Accelerated Virtualization

Storage Optimization

Total Customer Experience (TCE) & Manageability

• Bug fixes/ Hotfixes roll-up

• Seamless Integration into the Ecosystems

−

Failover Cluster aware storage pooling

−

Integrated with native management workflows (FoC

Mgr, SCVMM, Hyper-v Mgr, Server Mgr)

−

Supporting all native Hyper-V VM capabilities ( e.g.

Live/Quick Migration)

• Windows Server 2012 & R2 Support

−

Co-existence across Symantec storage management,

HA/DR

−

Hyper-V Virtual Fibre Channel support

SmartIO

• Read Cache based on SSD.

• Caching independent of SSD

vendor

• Online grow and shrink of

cache

AppHA for Hyper-V

• DG storage migration

• Hyper-V VM storage

migration

• Enclosure storage migration

• Storage Migration

performance Improvements

VBS support for FoC

Resilient Storage for Hyper-V

• VM specific storage services • Protect against array failures • Vendor independent Live