VMware Virtual SAN

Hyper-converged infrastructure software

Duncan Epping
Chief Technologist
Office of the CTO
Storage & Availability
Agenda

1 Introduction

2 Virtual SAN, what is it?

3 Virtual SAN, a bit of a deeper dive

4 Virtual SAN Recent Enhancements

5 Wrapping up

2
The Software Defined Data Center

• All infrastructure services virtualized:

compute, networking, storage
Management
• Underlying hardware abstracted,
resources are pooled

• Control of data center automated by

software (management, security)

• Virtual Machines are first class citizens

Compute Networking Storage of the SDDC

• Today’s session will focus on one

aspect of the SDDC - storage

3
Hardware
evolution
started the
storage
revolution
Simplicity: Operational / Management

5
The Hypervisor is the Strategic High Ground

VMware vSphere

x86 - HCI SAN/NAS Object Storage

Cloud Storage

6
Storage Policy-Based Management – App centric automation
Virtual Machine Storage policy
Reserve Capacity 40GB
Overview
Availability 2 Failures to tolerate • Intelligent placement
Read Cache 50%
• Fine control of services at VM level
Stripe Width 6
• Automation at scale through policy
• Need new services for VM?

vSphere • Change current policy on-the-fly

• Attach new policy on-the-fly
Storage Policy-Based Management

Virtual Datastore

Virtual SAN Virtual Volumes

7
Storage Policy Based Management – What does it look like?

If the storage can satisfy the VM

Storage Policy, the VM Summary tab
in the vSphere client will display the
VM as compliant.

If not, either due to failures, lack of

resources or other reasons, the VM
will be shown as non-compliant.

8
Virtual SAN,
what is it?

9
Virtual SAN, what is it?

Hyper-Converged Infrastructure

Software-Defined Storage

Distributed, Scale-out Architecture

Integrated with vSphere platform

vSphere & Virtual SAN

Ready for today’s vSphere use cases

10
But what does that really mean?

Exposing a single shared datastore

Virtual SAN

Leveraging local storage resources

VMware vSphere & Virtual SAN Integrated with your Hypervisor

Generic x86 hardware

VSAN network

11
VSAN is the Most Widely Adopted HCI Product

Simplicity is key, on an oil

platform there are no
virtualization, storage or network
admins. The infrastructure is
managed over a satellite link via
a centralized vCenter Server.
Reliability, availability and
predictability is key.

12
Virtual SAN Use Cases

Business End User

Critical Apps Computing DR / DA Test/Dev

DMZ Management Staging ROBO

VMware vSphere + Virtual SAN

13
Broadest Deployment Options from HCI to SDDC
Built on Industry-Leading VMware Hyper-Converged Software (HCS) EVO SDDC

vRealize

Certified Solutions Engineered Appliances NSX

VMware HCS VMware HCS VMware HCS

Virtual SAN + vSphere + vCenter Virtual SAN + vSphere + vCenter Virtual SAN + vSphere + vCenter

Lifecycle Management EVO SDDC Manager

EMC Federation
Virtual SAN Ready Nodes HCI Appliance
Certified Partner
Hardware

14
Tiered Hybrid vs All-Flash
Hybrid All-Flash

Caching
SSD PCIe Ultra DIMM SSD PCIe Ultra DIMM

Read and Write Cache Writes cached first,

Reads from capacity tier

Data
Virtual SAN Persistence
Capacity Tier
Capacity Tier Flash Devices
SAS / NL-SAS / SATA Reads go directly to capacity tier

100K IOPS per Host

40K IOPS per Host +
sub-millisecond latency

15
 Really Simple Setup

Manual or
Automatic Disk
Claiming
Deduplication
and
Compression
Enable?
Fault Domains,
2 node or
stretched
cluster?

16
 Virtual SAN,
a bit of a deeper dive

17
Virtual Machine as a set of Objects on VSAN

VM Home
• VM Home Namespace
VM Swap • VM Swap Object
VMDK • Virtual Disk (VMDK) Object

Snap delta • Snapshot (delta) Object

• Snapshot (delta) Memory Object
Snap memory
Snapshot

18
 Define a policy first…
Virtual SAN currently surfaces multiple storage capabilities to vCenter Server

What If APIs

New
Capabilities
in VSAN 6.2

19
 VMDK Object

Virtual SAN Objects and Components RAID-1

VSAN is an object store!

• Object Tree with Branches RAID-0 RAID-0

• Each Object has multiple Components

stripe-1a stripe-2a
– This allows you to meet availability and
stripe-1b
performance requirements stripe-2b

• Here is one example of “Distributed RAID” using Mirror Copy Mirror Copy
2 techniques:
– Striping (RAID-0) ESXi Host ESXi Host

– Mirroring (RAID-1)
• Data is distributed based on VM Storage Policy witness

ESXi Host

20
Number of Failures to Tolerate/Failure Tolerance Method
• Defines the number of hosts, disk or network failures a storage object can tolerate.
• RAID-1 Mirroring used when Failure Tolerance Method set to Performance (default).
• For “n” failures tolerated, “n+1” copies of the object are created and “2n+1” host contributing
storage are required!
esxi-01 esxi-02 esxi-03 esxi-04
~50% of I/O ~50% of I/O

RAID-1

vmdk vmdk witness

Virtual SAN Policy: “Number of failures to tolerate = 1”

21
Assign it to a new or existing VM
When the policy is selected, Virtual SAN uses it to place / distribute the VM to guarantee
availability and Performance

22
Fault Domains, increasing availability through rack awareness
• Create fault domains to increase availability
• 8 node cluster with 4 defined fault domains (2 nodes in each)

FD1 = esxi-01, esxi-02 FD3 = esxi-05, esxi-06

FD2 = esxi-03, esxi-04 FD4 = esxi-7, esxi-08

• To protect against one rack failure only 2 replicas are required and a witness across 3 failure domains!
FD1 FD2 FD3 FD4

RAID-1

esxi-01 esxi-03 esxi-05 esxi-07

esxi-02 esxi-04 esxi-06 esxi-08

vmdk vmdk witness

23
23
 Virtual SAN,
Recent Enhancements

24
Deduplication and Compression
RAID 5/6 support VSAN 6.2
Software Checksum March 2016
QoS via IOPS Limits
IPv6
Performance Service
Enhanced Capacity Views
Stretched Cluster
VSAN 6.1 Replication - 5 Minutes RPO
2-node ROBO
September 2015
Health Monitoring & Remediation

All Flash Configuration

64 node VSAN cluster
x2 Hybrid Performance
VSAN 6.0 VSAN Snapshots/Clones
March 2015 Health UI
Rack Awareness

VSAN 5.5
March 2014 25
 Today

Virtual SAN – Stretched Cluster

Active-Active data centers

witness
Site Recovery Manager

vSphere & Virtual SAN

• Virtual SAN cluster split across 2 sites!
• Each site is a Fault Domain (FD)
O
RP
in
• Site-level protection with zero data loss
vSphere
witness >5
m
e
nc

y
di
st
a
and near-instantaneous recovery
An

• Support for up to 5ms RTT latency

between data sites
– 10Gbps bandwidth expectation

• Witness VM can reside anywhere

– 200ms RTT latency

VMware vSphere & Virtual SAN – 100Mbps bandwidth required at most

vmdk vmdk • Automated failover

5ms RTT, 10GbE

26
Advanced Troubleshooting with VSAN Health Check

• Cluster Health
• Network Health
• Data Health
• Limits Health
• Physical Disk Health

• Stretched Cluster

• Proactive Tests

27
Virtual SAN 6.2
specifics

28
 in All Flash Only
w Beta
Ne 6.2

Deduplication and Compression for Space Efficiency

• Nearline deduplication and compression per disk group level.
– Enabled on a cluster level
– Deduplicated when de-staging from cache tier to capacity tier
– Fixed block length deduplication (4KB Blocks) esxi-01 esxi-02 esxi-03
• Compression after deduplication
– If block is compressed <= 2KB
– Otherwise full 4KB block is stored
vSphere & Virtual SAN

Significant space savings achievable,

making the economics of an all-flash
VSAN very attractive

vmdk vmdk
vmdk

29
 in All Flash Only
w
Ne 6.2

RAID-5/6 (Inline Erasure Coding)

• When Number of Failures to Tolerate = 1 and Failure Tolerance Method = Capacity  RAID-5
– 3+1 (4 host minimum)
– 1.33x overhead for RAID-5 instead of 2x compared to FTT=1 with RAID-1

• When Number of Failures to Tolerate = 2 and Failure Tolerance Method = Capacity  RAID-6
– 4+2 (6 host minimum)
– 1.5x overhead for for RAID-6 instead of 3x compared to FTT=2 with RAID-1

RAID-5

parity data data data

data parity data data

data data parity data

data data data parity

ESXi Host ESXi Host ESXi Host ESXi Host

30
 in
w
Ne 6.2

Software Checksum and disk scrubbing

Overview
• End-to-end checksum of the data to detect and resolve silent
disk errors due to faulty hardware/firmware
• Checksum is enabled by default (policy driven)
• If checksum verification fails on a read:
– VSAN fetches the data from another copy in RAID-1
– VSAN recreates the data from other components in RAID-5/6
stripe
Virtual SAN Datastore • Disk scrubbing is run in the background

Benefits
• Provide additional level of data integrity
• Automatic detection and resolution of silent disk errors
31
 in
w
Ne 6.2

Other new improvements

Client Cache
• Write through read memory cache
– 0.4% of total host memory, up to 1GB per host
• “Local” to the virtual Machine
• Low overhead, big impact!

Sparse Swap
• Reclaim Space used by memory swap
• Host advanced option enables setting policy for swap to
no space reservation
IOPS limit on object
• Policy driven capability
• Limit IOPS per VM/Virtual Disk
• Eliminate noisy neighbor issues
• Manage performance SLAs

32
 in
w
Ne 6.2

Enhanced Virtual SAN Management with New Health Service

Performance Monitoring Capacity Monitoring

Built-in performance monitoring

Health and performance APIs and SDK
Storage capacity reporting
And many more health checks…
33
 in
w
Ne 6.2

Performance, Scale and Availability for Any Application

BUSINESS-CRITICAL
A P P L I C AT I O N S

SAP Horizon Oracle

SAP Core Ready Tightly integrated cloud Oracle RAC supported

management
Testing and validated Testing and validated
deployments Bundles Virtual SAN deployments
licenses for lowest
cost VDI storage

34
Wrapping up

35
36
 o n
i si
V

VMware Virtual SAN: Generic Object Storage Platform

VMFS Block File Rest

Virtual SAN
VMware vSphere

37