#CiscoLiveAPJC © 2020 Cisco and/or its affiliates. All rights reserved.

Cisco Public
Introduction to
Hyperconverged
Richard Watson - DC Systems Engineer
BRKCOM-1110

#CiscoLiveAPJC
#CLMEL
#CiscoLiveAPJC BRKCOM-110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 3
Agenda
• Journey to Hyperconverged Infrastructure
• Introducing Cisco HyperFlex
• HyperFlex Data Platform
• HyperFlex DC & HyperFlex Edge
• Operations & Management
• HyperFlex Use Cases
• Summary

© 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public #CLMEL
#CiscoLiveAPJC BRKCOM-1110 4
Journey to Hyperconvergence

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 5
Server Storage Basics
• Simplified server overview
APP APP APP
• SAS or SATA drives connected to controller
OS (HBA)
• HDDs or SSDs
RAM CPU • RAID used to provide resiliency to drive
PCIe failures
NIC HBA • Presents “virtual drive” for file system
SAS/SATA • Controller connected to CPU over local bus
(PCIe)
• OS provides storage access through File
System to local apps
SSD HDD

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 6
 RAID Basics
Level Description Min. Disk Raw Disk Write
Disks Failures Util. Penalty

RAID 1 Disk Mirroring 2 1 1/2 2

A1 A1 (50%) (2 write)
B1 B1
C1 C1
D1 D1
D1 D2
RAID 5 Striped Blocks with 3 1 1 − 1/n 4
A1 A2 A3 Ap Single Distributed (3d, 66%) (2 read, 2
B1 B2 Bp B3 Parity write)
C1 Cp C2 C3
Dp D1 D2 D3
D1 D2 D3 D4
RAID 6 Striped Blocks with 4 2 1 − 2/n 6
A1 A2 A3 Ap Aq A4 Dual Distributed (4d, 50%) (3 read, 3
B1 B2 Bp Bq B3 B4 Parity write)
C1 Cp Cq C2 C3 C4
Dp Dq D1 D2 D3 D4
D1 D2 D3 D4 D5 D6

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 7
Shared Storage – Network • What about multiple
applications accessing same
data?
• Could share data over network
APP (CLIENT) • LAN or dedicated SAN
FILE SERVER APP (CLIENT)
OS
OS OS • LAN - Need a network enabled
file system (NFS, CIFS etc.)
Network RAM CPU
CPU RAM RAM CPU
• Storage server & network
PCIe

PCIe NIC HBA

PCIe

HBA NIC NIC HBA

becomes critical
SAS/SATA

SAS/SATA SAS/SATA
• Distance between App & Data
• Latency & Congestion will affect
app performance

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 8
 • Oversimplified!
Shared Storage – Resiliency • Dual storage controllers
• Separate servers
FILE SERVER MGMT FILE SERVER MGMT • Special RAID HBAs
OS OS
• Separate disk shelves
• Special “Dual-HBA” drives
RAM CPU CPU RAM

PCIe PCIe • Management software to ensure

NIC HBA HBA NIC
only one controller owns the
drives
SAS/SATA SAS/SATA

• Active/Standby
RAID • Scale with new shelves
Set
• Limited by compute
• Historically expensive & complex

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 9
 • Data storage demands explode
Shared Storage – Scale • Scale by adding new drive
shelves
FILE SERVER MGMT FILE SERVER MGMT • Eventually limited by storage
OS OS controller throughput
• Impacts all applications
RAM CPU CPU RAM performance
PCIe PCIe

• Only options to purchase

NIC HBA HBA NIC
another array
SAS/SATA SAS/SATA

RAID
Set

RAID
Set

RAID
Set

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 10
Parallel Distributed File Systems
New Approach Needed
• Google File System (GFS)
• Designed around 2003

• Built for Compute & Network limitations at the

time
APPS Controller APPS Controller APPS Controller

OS OS OS • Now retired

• Distributed controller/management service

RAM CPU RAM CPU RAM CPU
• Consistent file system across nodes
PCIe PCIe PCIe

NIC HBA NIC HBA NIC HBA • No central bottleneck

SAS/SATA SAS/SATA SAS/SATA
• No RAID required
• Replication across nodes built in to file system
itself

• Apps can read from local drive but writes must

be replicated and acknowledged first

Data Replicated over Network • Designed for multiple server failures

• Not initially designed for hypervisors

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 11
Hyperconverged Infrastructure (HCI)
Hypervisor & Virtual Machine Support
• Final piece for hyperconverged
infrastructure
• Different vendors do this differently!
Guest Guest Storage Guest Guest Storage Guest Guest Storage
VM VM Controller VM VM Controller VM VM Controller
• Distributed, parallel storage controller
HYPERVISOR OS HYPERVISOR OS HYPERVISOR OS
service

RAM CPU RAM CPU RAM CPU

• Proprietary file system
PCIe PCIe PCIe
• Owns the physical drives
NIC HBA NIC HBA NIC HBA

SAS/SATA SAS/SATA SAS/SATA

• May perform compression / deduplication

• Presents the hypervisor a supported file

system as if it were a NAS
• NFS, CIFS etc.

• VMs read/write to OS, which sends to local

Data Replicated over Network Storage Controller

• Abstracts complexity from hypervisor

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 12
Cisco HyperFlex

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 13
Approaches to Hyperconvergence
Vendor A - Dedicated Storage
Vendor A - Virtualised Storage

Vendor B - Compute Vendor B - Compute

Vendor C - Network Vendor C - Network

Traditional Infrastructure First Generation HCI

• Siloed Tiers of Compute, Storage and • Virtualise Storage • Still Separate Vendors
Networking from separate vendors combined with Compute
• Still Siloed Networking
• Siloed & Disaggregated Management • Storage and Disaggregated HW
Software Validated and SW management

Is this Hyperconverging?

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 14
Cisco’s Approach to Hyperconvergence
Vendor A - Dedicated Storage Cisco HX Data Platform

Cisco HCI
Vendor B - Compute
Cisco Fabric Computing
Vendor C - Network

Cisco Hybrid Cloud Mgmt.

Traditional
Infrastructure Cisco HyperFlex
• Multicloud Era • Cisco UCS Foundation
• Siloed Tiers of Compute, Hyperconverged Infrastructure • Converged Compute
Storage and Networking and Networking
• Fully Engineered Fabric Computing
from separate vendors
and Storage Data Platform with • Policy Based Single
• Siloed & Disaggregated Multicloud Service Integrations Management
Management
Complete Hyperconvergence

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 15
What is HyperFlex?
Fully Engineered and Optimized HCI

Performance and Scalability to Power

Any App, Any Cloud, Anywhere

HX Data Platform Hybrid Cloud Management UCS Foundation

• Purpose Built for HCI • On Premise & Cisco Intersight • Software Defined Compute
• Fast and Flexible • Programming/APIs • Fabric Centric
• Enterprise Grade Storage • Workload Optimisation • Intent-Based Policies
• Application Performance Management

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 16
HyperFlex Data Platform (HXDP)

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 17
The Cisco HyperFlex Data Platform
PROPRIETARY LOG STRUCTURED FILE SYSTEM
DESIGNED SPECIFICALLY FOR HYPERCONVERGENCE

ADVANCED FUTURE
DISTRIBUTED
DATA SERVICES READY

Object-Based File System Built Into File Designed for Containers

Architected for Scale-Out, System Architecture and Next-generation
Distributed Storage Applications

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 18
 HX Data Platform Node – Basic Hardware
• UCS Hardware
What is NVMe?
• Intel CPU(s)
• LSI HBA (SAS & SATA) • NVMe is a new protocol and
command set for PCIe
• NO RAID HYPERVISOR OS connected storage
• Cisco VIC • NAND or new 3D XPoint
(aka Optane or DCPMM)
• Different drive types
RAM CPU(s) • 65,535 queues vs.1
• OS Boot (M.2 NVMe)
• HX Drives PCIe Bus • Higher IOPs, Lower Latency
Converged
• Log Drive Ethernet (DCE) M.2 • Less CPU cycles per IO
VIC HBA
• Capacity Drives (6 min) (Boot)

• Typically SAS or SATA SAS/SATA Bus

• Cache Drive
• SAS or NVMe

• Multiple Hypervisors
• ESXi, Hyper-V, KVM*
HXDP LOG
`
HXDP CAPACITY HXDP CACHE
DRIVE DRIVES (6+) DRIVE

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 19
 HX Data Platform Node – Device Ownership
• Hypervisor OS installed to boot drive
HXDP Storage
• HXDP Storage Controller VM Controller VM
deployed
HYPERVISOR OS
• HCI not Magic, needs CPU/RAM
• PCI hypervisor passthrough
• DirectPath I/O
RAM CPU(s)

• Storage Controller VM owns: PCIe Bus

• HBA Converged
Ethernet (DCE) M.2
• ALL SAS/SATA Drives VIC HBA
(Boot)
• NVMe Cache Drive SAS/SATA Bus

• Hypervisor owns:
• M.2 Boot Drive
• VIC
• Other ( i.e. GPUs)
HXDP LOG
`
HXDP CAPACITY HXDP CACHE
DRIVE DRIVES (6+) DRIVE

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 20
 HyperFlex Data Platform Cluster
Hyperconverged Optimised, Scale Out and Distributed File System

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

HYPERVISOR CONTROLLER HYPERVISOR CONTROLLER HYPERVISOR CONTROLLER HYPERVISOR CONTROLLER HYPERVISOR CONTROLLER

CISCO
CISCO HYPERFLEX HYPERFLEX
DATA PLATFORMDATA
CISCO CISCO
PLATFORM
HYPERFLEX HYPERFLEX
DATA PLATFORMDATA PLATFORM

Start with as Few Hyperconverged Network Fabric Add Servers, One Distribute and Retire Older
as 2 (for Edge) or 3 Data Platform Policy Configures or More at a Time Rebalance Data Servers
(for DC) Nodes Installs in Minutes QoS Settings Across Servers
Automatically

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 21
 Continuous Data Optimisation
Inline Compression & Inline Deduplication

Log-Structured File System Yields More Efficient Data Optimisation

BEFORE AFTER

No Performance Impact
Inline Compression Inline Deduplication No Config lock-in
No Additional License

30–50% space savings 20–30% space savings Lower Cost

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 22
HXDP – Hypervisor Integration
• HXDP cluster presents one or more • For ESXi
virtual storage datastores to the • NFS (v3) used to mount HX Datastore
hypervisors • HXDP IOVisor kernel plugin required
• Acts as OS level NFS driver & proxy agent to Storage Controller VM cluster
• Where possible data services (i.e.
snapshots) offloaded to the • For Hyper-V
storage controller VM • SMB3 used to mount HX Datastore directly from Windows OS to the local Storage
• ESXi VAAI Controller VMs
• Windows OS provides native SMB3 failover cluster based resiliency.
• No IOVisor required

Guest Guest Guest Guest HXDP Storage Guest Guest Guest Guest HXDP Storage Guest Guest Guest Guest HXDP Storage
VM VM VM VM Controller VM VM VM VM VM Controller VM VM VM VM VM Controller VM

HYPERVISOR OS IOVisor VAAI HYPERVISOR OS IOVisor VAAI HYPERVISOR OS IOVisor VAAI

`
NFS/SMB3 NFS/SMB3 NFS/SMB3
HYPERFLEX DATASTORE/VOLUME

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 23
 Self-Healing Triggers
Disk Failure 1 min.
Non-Disruptive Operations Node Failure 2 hours

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

CONTROLLER CONTROLLER CONTROLLER CONTROLLER

HYPERVISOR VM HYPERVISOR VM HYPERVISOR VM HYPERVISOR VM

DATASTORE DATASTORE

File.vmdk A B C D E

A1 B2 C2 D3 E1 B1 A2 C3 E3 D1 C1 A3 D2 E2 B3 D1 E1 B3

• Stripe blocks of a file across servers • Restore back to original number of copies
• Replicate one or two additional copies to • Rebalance VMs and data post replacement
other servers (RF2, RF3) • Rolling software upgrades
• Handle entire server or disk failures

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 24
Data Protection and High Availability
Data Protected by Replication of Data Across the Cluster Nodes

Replication Factor 2 (RF2)

VM
Only Recommended for HX Edge (2-4 Nodes)
Write IO
Every block is written to 2 different nodes in the cluster

Lower availability to survive failures; Lower device

protection

Reduces raw disk capacity to 50% Replicate x1 x2

Disclaimer: While RF2 is available, RF3 is strongly recommended for high availability

Replication Factor 3 or 4 Node Cluster 5+ Node Cluster

Simultaneous Failures Supported: Simultaneous Failures Supported:
2
1 node / 1 drive 1 node / 1 drive

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 25
Data Protection and High Availability
Data Protected by Replication of Data Across the Cluster Nodes

Replication Factor 3 (RF3)

VM
Default and Recommended is Replication Factor = 3
Write IO
Every block is written to 3 different nodes in the cluster

Higher availability to survive multi-point failures; Higher

device protection

Reduces raw disk capacity to 33% Replicate x1 x2 x3

Note: RF3 is strongly recommended for high availability

Replication Factor 3 or 4 Node Cluster 5+ Node Cluster
Simultaneous Failures Supported: Simultaneous Failures Supported:
3
1 node / 2 drives* 2 nodes / 2 drives*
*drives across different nodes

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 26
 1. VM writes data to HX
datastore, received by the
Input/Output Lifecycles: Writes IOvisor, hashed to determine
the primary node for that
write, sent via network to
primary node’s controller VM.
2. Controller VM compresses the
data and commits the write to
VM IOvisor
CONTROLLER
IOvisor
CONTROLLER
IOvisor CONTROLLER the active local write log on the
VM VM VM
caching SSD, and sends
ESXi ESXi ESXi duplicate copies to the other
nodes.
Active Secondary Active Secondary Active Secondary 3. Additional nodes commit the
Local Write Log
Pair B1 A1 D1 C1 write to their passive write logs
on the caching SSDs
Passive Write Log A2 D2 B2 B3
Pair #1
C2 C3 A3 D3
4. After all 3 copies are
Passive Write Log
Pair #2
committed, the write is
Cache Drive Cache Drive Cache Drive acknowledged back to the VM
via the IOvisor and the
datastore like a normal I/O.
5. Additional write I/O is handled
Capacity Drive(s) Capacity Drive(s) Capacity Drive(s) the same way, until the active
local write log segment of a
Node 1 Node 2 Node 3 node becomes full, then a
destage operation is started.

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 27
 6. The active and
secondary log segments
Input/Output Lifecycles: Destage are flipped, the full
segments are ready to
destage meanwhile the
empty segments receive
new incoming writes.
IOvisor
CONTROLLER
VM
IOvisor
CONTROLLER
VM
IOvisor CONTROLLER
VM
7. The data is
deduplicated, and new
ESXi ESXi ESXi blocks are committed to
the capacity layer disks.
Active Secondary Active Secondary Active Secondary Duplicates generate only
Local Write Log
Pair B1 A1
E2 D1 A1 D1 C1 a metadata update
A2
E1 D2 A2 D2 B2 B3
Passive Write Log
Pair #1 8. Once all data has been
Passive Write Log C2 C3 E3
A3 D3 A3 D3 destaged, the secondary
Pair #2
Cache Drive Cache Drive Cache Drive
log segments are
emptied, and ready to
flip once the active
A2 D2 A1 D1 A3 D3
segments become full
Capacity Drive(s) Capacity Drive(s) Capacity Drive(s) again
Node 1 Node 2 Node 3

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 28
 Input/Output: Read Operations
• All systems have a small amount of RAM in the
VM IOvisor L1 Cache
controller VMs set aside for a read cache (L1).
CONTROLLER
VM
(DRAM) • All Flash read straight from capacity SSDs
ESXi
• Hybrid nodes have a large dedicated read cache segment
on their caching SSD (L2).
Active Secondary
Local Write Log
B1
• During destage, a copy of the most recently written data
Pair is also copied to the L2 read cache.
Passive Write Log E1
Pair #1
C2
• Requests to read data are also received by the IOvisor
Passive Write Log
Pair #2 and serviced in the following order:
Read Cache (L2) A2 D2 1. Active write log of primary node
(Hybrid ONLY)
2. Passive write logs
3. L1 (DRAM) cache
Cache Drive
4. L2 read cache (hybrid only)
A2 D2 5. Capacity SSDs/HDDs
Capacity Drive(s) • Reads are decompressed and cached in L1/L2 as
appropriate and returned to the requesting VM.
Node 1

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 29
Why HCI Software Matters…
IOPS in a 70/30 Workload – 1 Hour Duration

Source: https://research.esg-global.com/reportaction/ciscohyperflexcomplexworkloads/Toc
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 30
HyperFlex DC & HyperFlex Edge
- UCS Compute Foundation

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 31
Cisco HyperFlex Anywhere

HyperFlex DC Hybrid Cloud Stacks

DC Cloud
Any App, Any Scale Consistent On-prem + Cloud
Environment

Edge

HyperFlex Edge
Ultra-Light
Cloud Managed

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 32
HyperFlex DC – Fabric Networking

Mgmt
(OOB)
LAN SAN

• UCS Manager • Multiple HX Clusters supported on single FI pair

• Single-wire management
• Optional 16/32G Fibre Channel Uplinks to SAN – No
• Lossless UCS Fabric QoS Additional Cost
• Options for 25G or 40G Host Ports Fabric
Interconnects
• 40G or 100G Uplinks to LAN

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 33
 HXDP Storage
Controller VM

Volume Management
Device (VMD)
All NVMe RAM
HYPERVISOR OS

CPU(s)

PCIe Bus
Converged
Ethernet (DCE)
M.2

•
VIC (Boot)

NVMe Optane Cache

HXAF220C-M5SN • NVMe SSD Capacity
• 6-8 SFF NVMe Capacity Drives • NO SAS/SATA HBA! ` `
• 1TB -> 8TB NVMe SSD HXDP LOG
DRIVE
HXDP CAPACITY
DRIVES (6+)
HXDP CACHE
DRIVE

All Flash
PERFORMANCE

• SAS, NVMe or Optane SSD

Cache
• SAS SSD Capacity

HXAF220C-M5SX HXAF240C-M5SX
• 6-8 SFF 6G SATA Capacity Drives • 6-23 SFF 6G SATA Capacity Drives
• 960GB -> 7.6TB SSD • 960GB -> 7.6TB SSD

HX Edge
•
Hybrid •
SSD* SAS Cache
SAS HDD Capacity

HX220C-M5SX HX240C-M5SX HX240C-M5L

• 6-8 SFF 12G SAS Capacity Drives • 6-23 SFF 12G SAS Capacity Drives • 6-12 LFF 12G SAS Capacity Drives
• 1.2TB -> 2.4TB 10k RPM HDD • 1.2TB -> 2.4TB 10k RPM HDD • 6TB –> 12TB 7.2k RPM HDD
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 34
HyperFlex All Flash vs. HyperFlex All NVMe

70% more IOPS, 40% Lower Latency

Mixed VM
workload

ESG Report: https://research.esg-global.com/reportaction/ciscohyperflexallnvme/Toc?

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 35
HyperFlex Acceleration Engine

Software Hardware Accelerated

Data Path Data Path

I/O I/O • PCIe FPGA offloading HX filesystem compression

• Higher Compression, Lower $/GB
• Better VM Density, Lower $/VM
Comp/Decomp Driver
I/O
PCI Passthrough

Cache Cache Accelerator

vs. Card • Performance Boost + lower read/write latency
Dedupe Dedupe • CPU Utilization & Working Capacity Improvements
• Other service-offloads in-future: EC, Crypto, Hash etc.
Flush Flush

• Purpose Built & Co-Engineered Software & Hardware

Capacity Capacity
Controller VM Controller VM • Ongoing improvements via new images and driver
optimize
• Improved TCO

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 36
HyperFlex DC – Flexible Scaling
Mgmt
(OOB)
LAN SAN

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 37
HyperFlex DC – Compute Only Nodes
Mgmt
(OOB)
LAN SAN

• Only Unified Nodes require

HyperFlex licensing!
• Compute Nodes can utilise HX
storage at no cost
• Maximum 2:1 ratio of
compute to unified nodes

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 38
HyperFlex DC Scaling Limits
VMware HyperV
SFF
(excluding All-NVMe)
LFF All-NVMe SFF LFF
Converged Nodes Converged Nodes Converged Nodes Converged Nodes Converged Nodes
3-32 3-16 3-16 3-16 3-16

Compute-Only Nodes Compute-Only Nodes Compute-Only Nodes Compute-Only Nodes Compute-Only Nodes
0-32 0-32 0-16 0-16 0-16

✓Expansion Supported ✓Expansion Supported ✓Expansion Supported ✓Expansion Supported ✓Expansion Supported

2:1 2:1 2:1 1:1 1:1

Max ratio Compute to Max ratio Compute to Max ratio Compute to Max ratio Compute to Max ratio Compute to
Converged Nodes* Converged Nodes* Converged Nodes* Converged Nodes* Converged Nodes*

Max Cluster Size Max Cluster Size Max Cluster Size Max Cluster Size Max Cluster Size

64 48 48 32 32
! Hybrid Only ! Requires Enterprise License ! Hybrid Only
* 2:1 – Enterprise license (HXDP-P) if # of Compute > # of Converged Nodes.
* 1:1 – Standard license (HXDP-S) if # of Compute <= # of Converged Nodes.
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 39
What does the
future
look like? Traditional Infra IaaS/PaaS/SaaS
Public
Private Cloud
Cloud 1
“By 2022, more than 50%
Scale-Out Public Cloud
of enterprise-generated Infra
Core Cloud 2
data will be created and
processed outside [of] Mgmt. and Ops Public Cloud 3

the core data center or Edge Edge-Cloud

Apps and
cloud” Data Flows
IoT Branch/
12/18 Gartner, Real Time Remote
Gartner IT Infrastructure, Operations & Cloud Analytics
Strategies Conference Event Presentation, The
and
Future of IT Infrastructures: Always On, Always
Available, Everywhere, David J. Cappuccio, Bob Inference
Gill, et al., 3-6 December 2018

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 40
HyperFlex Edge Enabling Mass-scale ROBO
The Power of Cloud-Managed Computing
Ultra-light 2-node
HyperFlex Edge
Lights out, multi-site
parallel deployment to
1000’s nodes

Customer Leveraging
Rapid cluster profile
Existing Branch Network
creation and full
stack upgrades

Flexible scalability
and investment WAN
protection

Cloud Witness Service

Connected TAC
Support Three-node cluster quorum
without physical 3rd node!

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or

© 2019 Ciscoits affiliates.
and/or All rights
its affiliates. All reserved. CiscoCisco
rights reserved. PublicPublic 41
Benefits of Invisible Cloud Witness Architecture
Traditional 2-Node Cisco HX 2-Node • No additional license cost; Included in an HX Edge
Deployment Deployment Subscription
2-Node ROBO 2-Node ROBO 2-Node ROBO • No requirement for 3rd site or existing infrastructure
Site #1 Site #1 … Site #2000

VM VM VM VM VM VM
• Cloud-like operations - Nothing to manage
• No user interface to monitor
VM VM VM VM VM VM
Hypervisor Hypervisor Hypervisor
• No user driven software updates
• No setup, configuration or backup required
• No scale limitations

• Security
• Realtime updates with the latest security patches
WAN WAN • All communications encrypted via TLS 1.2
• Uses standard HTTPS port 443; no firewall configuration
required

• Built on efficient protocol stack

• No periodic heartbeats
3rd Site for • No cluster metadata or user data transferred to Intersight
Witness VMs • Tolerates high latency and lossy WAN connections
$$ CapEx & OpEx $$

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 42
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 43
HyperFlex Operations and Management

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 44
HX Connect – Local Cluster Management
Monitor from Robust, Smart, Insightful Secure
Anywhere, Anytime Extensible Analytics Management
Interface
• Intuitive and Complete HX • Hypervisor Agnostic • Monitor Performance and • Discrete Read-only and Sys-
Monitoring and Management Capacity in Admin RBAC
Management Real-time
• Built-in Automation with • 100% vCenter and SCVMM
• Ready for Mobile RESTful API • Plan Future Needs Based on Integrated Administrative
Monitoring and Historical Trends Roles
Management

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 45
HyperFlex Connect
Dashboard

• Easy to read Dashboard,

focusing on operational
visibility and
management

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 46
HyperFlex Connect ANALYZE: Detailed cluster
performance analytics

Performance
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 47
HyperFlex Connect
MANAGE: Datastore Life-Cycle
Management, Create Ready
Clones and Cluster Upgrade

Datastores

Ready Clones

Upgrade
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 48
One-Click Full Stack Hyperconverged Cluster Upgrades

HCI vendors use different management tools to With Cisco HyperFlex, one-click, fully integrated
upgrade each portion of the HCI stack in a cluster. stack upgrades are possible via HX Connect.*

Storage Stack Storage Manager HXDP Storage Stack

Hypervisor OS Hypervisor Manager
HX Connect ESXi Hypervisor OS

OTHER HCI
Server Firmware Server Manager UCS Server Firmware

VENDORS
Storage Stack
Hypervisor OS
Storage Manager

Hypervisor Manager
HXDP Storage Stack
ESXi Hypervisor OS
Server Firmware Server Manager UCS Server Firmware

COMPLEX!
Storage Stack
Hypervisor OS
Storage Manager

Hypervisor Manager
HXDP Storage Stack
ESXi Hypervisor OS
Server Firmware Sever Manager
Non-Disruptive UCS Server Firmware
online upgrades!
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 49
 Localisation
• HX Connect is localised to
Chinese, Japanese and
Korean.
• The UI auto-detects the
locale, based on the
browser locale + language
and switches to the
translated version.

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 50
REST APIs
• Well-documented syntax and
examples with REST API explorer
• Secure token-based access with
RBAC and auditing
• Eco-system: UCS Mgmt., Services
(CMS), 3rd party (Veeam, etc.) Accessed Via: http://<Cluster-IP>/apiexplorer
API example
• Expanded coverage by adding AAA,
Encryption, Data Protection,
Support, and Performance APIs

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 51
HX Sizing Tools

Sizer Workload Profiler HX Bench

• End-to-End Sizing • OVA deployed in existing • OVA with benchmarking
• Includes – compute (CPU & environments built on vdbench
Memory), Storage • Quantify usage for sizing • Benchmarking made easy
performance and capacity • Integration with sizer to • Follows industry standard
• Application templates to aid automate end to end sizing benchmarking practices for
application based sizing of environments a realistic readout

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 52
Hypercheck : Hyperflex Health & Pre-Upgrade Check
Tool
• To be run before upgrade
procedure
• Supported HX Systems
• 2.6, 3.0, 3.5, 4.0
• ESXi
• Hyperflex Standard Cluster
• Hyperflex Edge Cluster (3
Node & 4 Node)

https://www.cisco.com/c/en/us/support/docs/hyperconverged-infrastructure/hyperflex-hx-data-
platform/214101-hypercheck-hyperflex-health-pre-upgr.html

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 53
 HyperFlex Data Protection
Flexibility to meet business needs
Integration with 3rd Party Backup Vendors
> 24 hours

Backup & DR
to Cloud

In-built Snapshots, Replication & Disaster Recovery

RPO / RTO
5mins – 24 hours

High Performing VM VM
VM Centric Planned Migration,
Periodic Snapshots VM VM Disaster Recovery and
Test Recovery
VM Highly optimized long distance VM
Async Replication

Local High Availability and Stretched Clusters

Zero

Synchronous

Replication

Local Metro WAN Cloud

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 54
Logical Availability Zones (LAZ)
Cluster Scale With High Availability
• Increased resiliency without added
manageability overhead HX DATA PLATFORM
• How does it work? RF3
• HX nodes grouped into logical
“availability groups”
(N/A for compute nodes)
• HXDP never places 2 copies of the data
in the same availability group
• Clusters with LAZ can survive > 2 Availability Availability Availability
simultaneous node failures without Group 1 Group 2 Group 3
data loss or loss of availability
• Tolerate more independent failures
• Can be used for ensure Rack/Room
resiliency

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 55
HyperFlex Stretched Cluster
Cloud Scale Data Platform

Power Mission Critical Apps with

Disaster Avoidance Maximum Uptime Zero RPO Automated DR

DB APP
RF2+2 APP DB

HX Data Platform

Synchronous Replication

10Gbps /5ms RTT

Site-A Site-B
SSD SSD SSD SSD SSD SSD

Witness Server
3rd site
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 56
HX Stretched Cluster with VXLAN EVPN
Ultimate Flexibility - Any App, at Any DC, at Anytime
VXLAN DCI VXLAN
SPINE SPINE

/32 Host DCI /32 Host

Route for Route for
DB VM App VM

WAN
VXLAN VXLAN
LEAF LEAF

Synchronous Replication
Bare Metal Bare Metal
L4-L7 Services L4-L7 Services
Servers 10Gbps /5ms RTT Servers

DC1 DC2
DB APP

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 57
 HyperFlex Data Protection
Built-In 1-Click Disaster Recovery

App Web
App Web

DB DB

HX Data Platform HX Data Platform

Long Distance Replication

Test Recovery Planned Migration Unplanned Failover

• DR Readiness • Move VMs across Data Centers • Recovery VMs after Disaster
/ Clusters • Support for VMWare Site
• Customise DR Test parameters Recovery Manager (SRM)
• Re-Protect after Migration • VMWare certified Storage
• Powershell Runbooks Replication Adapter(SRA)
• Re-Protect after Recovery

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 58
HyperFlex Data Protection
Protect your Entire Infrastructure with 3rd Party Backup Products

VMware UCS
Cisco HyperFlex
HyperFlex Platform
VMWare
Snapshot Snapshot
APIs APIs

SAN
FI pair

Backup Async WAN data

replication
3rd Party Backup Vendor
3rd Party Backup Vendor DR Site Media Server
Primary Site Media Server (UCS (UCS S3260/240)
S3260/240)

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 59
Cisco Intersight &
HyperFlex

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 60
Cisco Intersight 450,000+
Intelligent Centralised Management Connected Servers
SaaS Actionable
Simplicity Intelligence 74,000+
Intersight
Connected
Fabric Interconnects
74,000+
Connected
Fabric Interconnects

Unified Management Recommendation Engine Enhanced Support Experience

Embedded recommendation platform Hosted platform allows Cisco to address
Single pane of glass, consistent operations
with insights sourced from across Cisco issues platform-wide and experience
model and experience for managing all
installed base and tailored to each extends into TAC supported platforms
systems and solutions
customer

Policy Based Orchestration Programmability

Deploy services using intent-based, reusable, End to end programmability with native API,
consistent policies across all managed SDK’s and popular DevOps toolsets will enable
devices customers to consume natively

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 61
Intersight Dashboard

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 62
Centralised Management: HX Cluster Deployment
Intersight

• Deploy from anywhere!

• No need for existing infrastructure
or OVA-based installation
application
• Latest version always available
• Reusable policy for rapid &
consistent deployment
• Simple ramp-up of large HX projects
with simultaneous background
deployment

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 63
Telemetry & Analytics: HCL Check
Intersight

• Automatically validate against the

HCL
• Check adapter & BMC firmware and
drivers against the current OS
• View status in server list view

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 64
Telemetry & Analytics: Recommendation Engine

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 65
Simple Worldwide Infrastructure Upgrades
Powered by Cisco Intersight
• Intersight wizard enables optimized HyperFlex
data platform upgrades.
• Choose one or many HX Edge clusters to upgrade
in parallel

HX-ROBO
CHS
Upgrade

HX-ROBO
NYC

Non-Disruptive online upgrades!

HX-ROBO HX-ROBO
HX-ROBO
Beijing SFO
Denver
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 66
Intersight – Connected TAC
Data automatically Data analyzed,
collected problems identified

Website
/phone

Smart
TAC agent
Call Home
identifies next steps

Time

#CiscoLiveAPJC #CLUS © 2020 Cisco and/or 67 67

BRKCOM-1110 © 2019 Ciscoits affiliates.
and/or All rights
its affiliates. All reserved. CiscoCisco
rights reserved. PublicPublic BRKINI-2259
 NEW
Cisco Intersight Workload Optimizer
Ensure app performance
and critical digital
experiences

Understanding app resource needs at

every layer

AI-assisted analytics drive automated

resourcing decisions

Continuous optimization matches

application resource demands to the
underlying infrastructure

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 68
Common HyperFlex Use Cases

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 69
HyperFlex Performance for Mission-Critical Apps
Sustained Enterprise
Low Latency Data Services
Availability &
Resiliency

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 70
HyperFlex – Multiple Clusters, Separate Functions

Production DMZ Development

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 71
HyperFlex & Cisco Workload Optimiser
Workload • Workload Optimiser
Optimiser can logically merge
multiple heterogeneous
HyperFlex clusters to
optimize the entire
“Super-Cluster”
• Analysis accounts for
different versions of
HyperFlex (hybrid vs.
all-flash) resulting in
performance-based
All Flash Cluster Hybrid Cluster storage tiering.
“Super-Cluster”

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 72
3rd Party File Services on Cisco HyperFlex
• 3rd party SMB/CIFS file services
on Cisco HyperFlex
• Windows and Nexenta
• Home directory and VDI use
cases
• Best practices and reference
configurations for ease of
deployment

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 73
 Cisco Collaboration on Cisco HyperFlex
Wide Portfolio of Collaboration Applications
WebEx & Prime Collaboration Unified Unity Connection Instant Messaging &
Telepresence Communications Presence
Manager (CUCM)

Private, Hybrid or Partner Cloud deployment choices on HyperFlex

✓ Prescriptive Configuration Tested Reference ✓ Deterministic Capacity & Superior

✓ Architecture Best-practice Specs-Based Performance
Configuration (TRC)
✓ UC Sizing Tool Supported Ready Nodes ✓ Configure To Order
Certified ✓ Deployment Flexibility

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 74
Citrix VDI on Cisco HyperFlex
Citrix Cloud Cisco HX
XA/XD Service, ShareFile,
Lifecycle Mgmt, Secure Connect
Browser HX Connect, Multi-site
API Infrastructure Mgmt

End Users
Subscribed to
Infrastructure &
Workspace Services

Citrix Admin Cisco Admin

• Enterprise IT • Enterprise IT
• Service Provider • Hosted Provider
• Workspace Publisher Server/Desktop VDAs, AD Server/Desktop VDAs, AD Server/Desktop VDAs, AD Server/Desktop VDAs, AD • Infrastructure Services

Site 1 Site 2 Site 3 Site 4

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 75
HyperFlex GPU Use-Cases & Options
Virtual Desktop Infrastructure Artificial Intelligence/Machine
(VDI) Learning (AI/ML)
Accelerated Graphics for Training Inferencing
Virtual Desktops
x2 x2
x2 HX220
HX240
HX240 • NVIDIA Tesla V100
• AMD Firepro 7150x2 • NVIDIA Tesla V100-32 x6
• NVIDIA Tesla M10 HX240
• NVIDIA Tesla P40 • NVIDIA Tesla P4
• NVIDIA Tesla P100-12G
C480ML
• NVIDIA Tesla P100-16G x8
• NVIDIA Tesla M60 • NVIDIA SXM2 V100

Better Together, Customer Choice, Cisco Validated Design with Eco-system

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 76
HyperFlex Edge & SD-WAN Integration
3.
Fully automated & cloud managed branch-
vMANAGE in-a-box; compute, storage, switch fabric &
SD-WAN.

1.
WAN
From Cisco Intersight:
1. Auto-generate
2. 2. switch configurations
* via Intersight in Phase 2
2. Automate remote
deployment of virtual routers

2-Node ROBO 3. View SD-WAN

Customer Site
status & telemetry

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 77
Evolution from Virtualisation to Containerisation
Google Cloud’s Red Hat OpenShift Cisco Container
Kubernetes
Anthos Container Platform Platform
25%
of enterprises use
35% 29% containers in production
<18% Decline
Bare metal servers Bare metal servers

58%
<14% Decline 37% of IT Operations teams drive
43% Virtual machines container adoption
Virtual machines

52%
>54% Growth 34% Containers said their incumbent
22% Containers
networking vendor was “very
helpful” in supporting
% of production workloads run on each % of production workloads run on each container-based environments
server type today server type 24 months from now
Source: ESG “Trends in Modern Applications Environments”,
Aug-19 #CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 78
 HyperFlex Storage Integration for Kubernetes
Dynamic on-demand provisioning of Kubernetes persistent volumes through HyperFlex

Developer requests a persistent volume

1
through a persistent volume claim
Kubernetes Cluster

Kubernetes Worker Kubernetes Worker Kubernetes Master Persistent volume claim request is sent to the
HyperFlex Provisioner (pod running in
2
Kubernetes) via the StorageClass parameters
Pod1 Pod2 Pod3 HyperFlex
Provisioner
Pods Pods Pods

3 HyperFlex Provisioner makes call to

Persistent
Volume Claim
Persistent
Volume Claim
Persistent
Volume Claim
HyperFlex API to provision backend iSCSI LUN
for request
HyperFlex Plugin Kubelet HyperFlex Plugin Kubelet HyperFlex Plugin Kubelet
Software iSCSI initiator on each Kubernetes
Software iSCSI Initiator Software iSCSI Initiator Software iSCSI Initiator
4 node VM discovers new iSCSI block device
from HyperFlex cluster

iSCSI iSCSI iSCSI

LUNs LUNs LUNs
HyperFlex Plugin orchestrates the mounting
5
of the new iSCSI block device and the creation
HyperFlex API
of a Kubernetes persistent volume
Cisco HyperFlex
Kubernetes binds the newly created persistent
6
volume to the persistent volume claim

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 79
HyperFlex Application Platform (HXAP) for Kubernetes*
• Intersight based Full-Stack Management, Monitoring &
CCP Telemetry.
Intersight SaaS or • Cisco Container Platform (CCP) run from Intersight
On-prem • Managed Multi-Tenant Kubernetes
• K8S Work/Master VMs deployed from Intersight
• Modern KVM Hypervisor
• HX Data Platform with K8S Storage Integration
K8s nodes
• No V-Tax!
K8s nodes HXDP
K8s nodes
K8s Nodes Controller
HXAP
Infra VM
• Enterprise grade solution
VM
• Host clustering
Linux / KVM Hypervisor • Resource scheduling
• Orchestration,
• HA, Live Migration etc.
• Extensible in future to run General Purpose VMs, Bare
Metal Container Hosts & Serverless**
HyperFlex Application * Ph1 Beta: Q1 CY20, GA Q2 CY20
Platform (HXAP) ** General purpose VMs – 2H 2020
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 80
 Are you ready
Use this slide for transitions
for the future today?
• …or just keeping the lights on?

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 81
HyperFlex -Your Applications, Anywhere
Enterprise Apps

Hybrid/Multicloud
Cisco Cloud
Center

Containers / Cloud Native Apps

Core Cloud
Cisco Container
Google
Platform
Anthos

Intelligent Management (AI Ops)

Cisco Cisco Workload

Edge Intersight Optimization Manger

AI/ML Apps

Secondary Storage / Backup and Recovery

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 82
HyperFlex Market Momentum

Leader Overall Winner

Gartner MQ for HCI HCI Product of the Year
Nov, 2018 Dec, 2019

Leader Leader
Forrester Wave for HCI Gartner MQ for HCI
Oct, 2018 Nov, 2019

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 83
 Cisco HyperFlex – Built for the Future
Complete Hyperconvergence with Game Changing Results Today!

Server Databases Virtual Desktop Containers ROBO & Test and

Virtualization & ERP Infrastructure IoT Edge Development

4000+ Customers Are Saying:

3x Lower Latency vs
Competitive HCI 1 50% Improvement in ERP
Response Time* 5 30%More Performance for
Mission Critical Apps* 91% Downtime Reduction 1

50% Lower Cost of

Operations 2 80% Savings vs. 3-tier
Infrastructure* 1 75% Management
time savings* 51% Savings vs. Public Cloud
Over 3 years* 2

1: 2019 ESG, Mission-critical Workload Performance Testing of Cisco HyperFlex All-NVMe with
Intel Optane SSD on the Cisco Unified Computing System Platform (UCS)
2: 2019, IDC Business Value of Improved Performance and Agility with Cisco HyperFlex
*See Cisco Sales Representative for more information #CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 84
Thank you

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 85
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 86
HyperFlex Solutions – CVDs & White Papers
Virtual Server Infrastructure Virtual Desktop Infrastructure
• CVD: HyperFlex 3.5 Stretched Cluster for VSI with
vSphere 6.5 and ACI 4.0 Multi-Pod Fabric
• CVD: HyperFlex 4.0 Edge with Intersight
Recently • WP: Performance Testing on HyperFlex 4.0 with All
NVMe
Published
• CVD: HyperFlex with Cohesity Data Protection
• CVD: HyperFlex Edge Protection with Veeam
• CVD: HyperFlex 4.0 for VSI with VMware ESXi

• WP: Performance with Intel 2nd Gen Xeon Scalable • WP: Horizon 7.x on ESX Infra on HX240 M5, B200 M5
Processors Graphics Performance
• WP: Adaptive Infrastructure with HyperFlex and FC • WP: Citrix 7.x on ESX Infra on HX240 M5, B200 M5
Coming Soon shared storage Graphics Performance
• CVD: Cohesity Data Platform with Cisco UCS S3260 • CVD: Citrix 7.x on Hyper-V 4000 Seat HyperFlex
Storage Servers (including HX Edge) 4.0(1b) on HXAF 220 M5

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 87
HyperFlex Solutions – CVDs & White Papers
MS SQL / Hyper-V / Containers Other Workloads on HX

• WP: SAP HANA on HXAF with Cascade Lake

• CVD: MS SQL 2019 on HX with All NVMe and ESXi
• WP: Oracle DB on HyperFlex with All NVMe
• CVD: HX 3.5 AF for SQL 2016 with ESXi
Recently • CVD: HX 3.5 AF for SQL 2016 with Hyper-V • WP: Oracle RAC on HyperFlex with All NVMe

Published • CVD: HyperFlex 4.0 for VSI with Hyper-V • CVD: Splunk on HyperFlex
• WP: Google Cloud Anthos 1.0 on HX Reference • WP: SMB File Services on HX with CTERA
Architecture
• WP: SMB File Services on HX with Nexenta

• WP: Google Anthos 1.2 with HX CSI

Coming Soon • WP: Epic end-to-end deployment on HyperFlex
• WP: SQL 2019 Big Data Cluster Ref Arch on HX

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 88
HyperFlex Stretched Cluster
Where can I find more information?

Stretched Cluster White Paper :

https://www.cisco.com/c/dam/en/us/products/collateral/hy
perconverged-infrastructure/hyperflex-hx-
series/operating-hyperflex.pdf

Stretched Cluster Install Guide :

https://www.cisco.com/c/en/us/td/docs/hyperconverged_s
ystems/HyperFlex_HX_DataPlatformSoftware/Installation
_VMWare_ESXi/4_0/b_HyperFlexSystems_Installation_
Guide_for_VMware_ESXi_4_0.html

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 89
IDC Report HyperFlex Customer Results
Lower
Cost 452% five-year ROI
50% lower cost of operations
8 months payback

Increased
Agility 91% less unplanned downtime
50% faster development life cycle, new
features

More
Efficiency 71% more efficient IT infrastructure teams
93% less staff time to deploy new servers

Read the full report: http://cs.co/HyperFlexIDC

#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 90
#CiscoLiveAPJC BRKCOM-1110 © 2020 Cisco and/or its affiliates. All rights reserved. Cisco Public 91