How to Troubleshoot Apps for the Modern Connected Worker
Cisco HyperFlex 3.0
1. Jonathan Gorlin, Aaron Kapacinskas, Robert Quimbey, Michael Zimmerman
HyperFlex Product Team
May 24, 2018
Any App. Any Cloud. Any Scale.
Cisco HyperFlex 3.0
9. Microsoft Hyper-V Support
Integrated Management
• Manage Hyper-V hosts from familiar SCVMM, Hyper-V manager,
and PowerShell
• Automate HX operations using HX REST API
• Leverage HX Connect UI (HTML5) for HX management
Enterprise workloads, Validated solutions
• Citrix VDI deployment (XenDesktop and XenApp)
• Microsoft SQL, Exchange, Sharepoint etc.
• Oracle & other mission critical apps
• Backup vendor integrations
Industry Leading Foundation
• Log Structured Filesystem & Data services designed for HCI
• Highly scalable, resilient, scale-out SMB3 file server protocol
• Fully Data distributed, No Data locality
Windows Server with Hyper-V
• Windows Server 2016 Datacenter
• HX220 M5 and HX240 M5: Hybrid-Flash & All-Flash
• Microsoft native failover clustering, checkpoint (snapshots), replica
support, AD support
CRTL
VM
CRTL
VM
CRTL
VM
VM VM VM VM VM VM VM VM VM
HyperFlex Data Platform
10. Hyperconverged Scale Out and
Distributed File System
Start with as Few
as Three Nodes
Hyperconverged
Data Platform
Installs in
Minutes
Network Fabric
Policy Configures
QoS Settings
HYPERCONVERGED DATA PLATFORM
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
11. Hyperconverged Scale Out and
Distributed File System
Start with as Few
as Three Nodes
Hyperconverged
Data Platform
Installs in
Minutes
Add Servers,
One or More at a
Time
Network Fabric
Policy Configures
QoS Settings
Distribute and
Rebalance Data
Across Servers
Automatically
Retire Older
Servers
HYPERCONVERGED DATA PLATFORM
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
VM VM VM
HYPERVISOR CONTROLLER
HYPERCONVERGED DATA PLATFORMHYPERCONVERGED DATA PLATFORM
12. DATASTORE DATASTORE
• Stripe blocks of a file across servers
• Replicate one or two additional copies to
other servers
• Handle entire server or disk failures
• Restore back to original number of
copies
• Rebalance VMs and data post
replacement
• Rolling software upgrades
Non-Disruptive Operations
CONTROLLER
VMHYPERVISOR
VM VM VM
CONTROLLER
VMHYPERVISOR
VM VM VM
CONTROLLER
VMHYPERVISOR
VM VM VM
CONTROLLER
VMHYPERVISOR
VM VM VM
File.vmdk
D1 E1A1 B1 C1B2 A2 A3C2 C3 D2D3 E2E3 D1E1 B3 B3
EDCBA
13. Capacity and Network Utilization
DATASTORE
HYPERVISOR
1 2 3
Balanced Space Utilization
No Data Migration on VM Migration
Less Stress on Network
• HX balances space utilization: no data migration required following a VM migration
CONTROLLER HYPERVISOR CONTROLLER HYPERVISOR CONTROLLER
VM VM VM
14. Continuous Data Optimization
BEFORE
InlineDeduplication
20–30% space savings
InlineCompression
30–50% space savings
No Special Hardware
No Performance Impact
No Config lock-in
No Additional License
Log-Structured File System Yields More Efficient Data Optimization
Lower Cost
15. ReadyClone PowerShell Script
• Declare source VM, Clone Name, and # of Clones
• HyperFlex Clone is nearly instantaneous, 2-5s per clone due to
registering it
in Hyper-V
Powershell ReadyClone
17. Download Remote Server Admin tools for Windows 10 to remotely manage
Server Core
• Hyper-V Manager
• Management at local Hypervisor granularity
• SCVMM
• Management at cluster and beyond
• PowerShell
• All Windows Management has extensions in Powershell
• Failover Cluster Manager
• Enables VM failover and Hypervisor Clustering
• STCLI / HX Connect
• Manager Storage Cluster, services, and hardware
Management Options
22. Production VM Checkpoint – It just works
• WMI in Host, VSS in Guest (Linux too!)
• No longer snapshots the Host and then the Guest
• Uses differential avhdx disk for new writes.
• Many backup applications will remove the checkpoint after backup complete.
• Avhdx files consolidated into vhdx files.
• RCT & MRT files persist CBT for both live migration and host crash
Windows 2016 Production Checkpoint
23. Data Protection with Hyper-V
1 Add SCVMM | HX Cluster 1
Create backup job & select HX1
Host & VM SQL 1
2
3
Windows Server 2016
Production Checkpoint via WMI
4
VSS writer in “Guest
VM” SQL1 gets
invoked.
5
For VHDX sql-share-1
• aVHDX (child file) 1 gets
created
• PrimaryVHDX becomes
read-only
• All rights go into aVHDX till
the backup job is completed
6
For subsequent
backups, backup
vendor leverages
Resilient Change
Tracking by Hyper-V
Cisco C240 M5
Backup
Server
Backup
Repository
Cloud
Connect
Availability
Console
Veeam Availability Suite
Backup & Restore 9.5
SCVMM | HyperFlex Cluster 1
HX1
HX2
HX3
HX4
HX5
HX6
VM SQL 1
VHDX
sql-share-1
25. What is a Stretched Cluster?
VM VM VM VM
VM
VM VM VM VM
VM
Single Cluster
3rd site
Witness Server
1. A stretched cluster is a single cluster with nodes geographically
distributed
2. Storage is mirrored across each sites
1. Sites need to be connected overlow latencynetwork
4. Geo-failover (VM) is like failover in a single cluster
5. “Split-Brain”:Conditionwhen nodes on either sites cannot see
each other
• Network failure
• Site failure
6. “Witness” : An entityhosted on a 3rd site responsible for
decidingthe which site survive after a split-brain
26. HyperFlex Stretched Cluster Use Case
Active-Active Disaster Recovery solution for Mission Critical workloads with
extremely high uptime (near Zero RTO) and no data loss(Zero RPO)
• The DR site could be a data-center across the street
• The DR site could be few 100 kms apart
• No need for a DR orchestration tool
HX Data Platform
App DB App DB
Site-A Site-B
Synchronous Replication
27. HyperFlex Stretched Cluster
ZERO RPO, NEAR ZERO RTO
Site-B
VM VM VM VM
VM VM VM VM
Site-A
3rd site
Configuration Support
Single Stretched Cluster across
2 sites
Symmetric Configuration
Site to host a “Witness Server”
(small VM)
8 +8 nodes on each site
IO Path
Active-Active sites – VMs
Active on each site
VM Read IOs served locally
VM Write IOs Sync-Writes
across sites
2x copies on each site
HA Operations
Recover from a Site failure
Recover from a Local failure
Failover of VM
vMotion of VM
Split Brain handling
Management
Cross site Cluster creation
Non disruptive online rolling
upgrade
Site awareness in HX Connect
Site specific Alarm and Events
on a single Dashboard
Witness Server
29. 1 Pair SitesConfigure Network Create Protection Group
Create a Replication VLAN
Input the network parameters
Set the Replicationbandwidth
One time setup
Assign a name
Provide remote site credentials
Map Datastores
Supports a single site pairing
Create one of more Protection Groups
Provide a replicationinterval
Select VM consistent
2 3
HyperFlex Disaster Recovery
Setup
30. HyperFlex Replication & Disaster Recovery
Overview
Point-to-Point(1-to-1)Replication
Different number of Cluster nodes
on each site
Replicate between HX220 and
HX240 Clusters
Replicate between AFand Hybrid
Supported Configuration
Recover from the Snapshot copy
Supports plannedand unplanned
recovery
Active – Active and Active-Passive DR via
bi-directional replication
Support for Failover, Fast Failback& Test
Recovery
Underlying Technology Disaster Recovery
Snapshot based, PeriodicReplication
Scale-out,performant,reliable and network
optimized
VM CentricReplication
PIT (latest)image for Recovery
Flexible RPO of 5mins to 24 hours
HTML 5 based , REST API & CLI based
operations and monitoring
DB
WebApp
Long Distance Replication
on WAN
HX Data Platform
DB
App
HX Data Platform
Business
Service
Business
Service
FailoverWeb
31. VMWare
Identify failover network
requirements (vSwitch,vLAN,
PortGroups, IPs, DNS)
Identify VM or Protection Groups
Prioritize VMs based on RPO & WAN
considerations
VMs load order
Replication of cloned VMs will take
up more space at destination than
at source.
Network
Use the HX Maximum
Bandwidth setting to
Limit use of scarce WAN bandwidth
Limit perf impact on primary I/O of
the two clusters
Use a secure tunnel over the
WAN if concerned about data
security.
MTU is 1500 and cannot be
changed through the UI.
HX Models
You can mix clusters with HX240 &
HX220 but be aware of capacity
limits
Number of nodes and type can be
different on each site ( All-Flash to
Hybrid replication)
For optimal usage of resources
consider Active-Active Cluster
Need to plan for the storage
capacity based on the number of
VMs to be protected
HyperFlex Disaster Recovery
Checklist
33. Elastic Scaling Options in HX 3.0
Scale with availability
Without compromising simplicity
Up to 64 Node clusters
Hybrid & All Flash
Flexible drive capacity options
LFF HDD Support
HX 3.0
34. Intel Optane Support
I N T E L O P T A N E S U P P O R T
• Caching drive option available with
HX 3.0
• Available on M5 only
• All Flash only configurations
• Part number: HX-NVMEXP-I375
• 375G drive capacity, 3D
XPoint™
• Advantages:
• Higher Endurance: 20.5PBW (~
30DWPD)
• Higher drive level performance
HX 3.0
Intel® Optane™ SSD DC P4800X Series
(Formerly known as Cold Stream)
(Specs)
35. HX Capacity Scaling Options
1 . 8 T B H D D O P T I O N L F F S U P P O R T
• 1.8TB 10Krpm SFF HDD
• Lower overall $/GB, similar performance
• Support for M4 & M5
• Supported in 220 and 240 models
• For high capacity use cases, lower $/GB
• M5 240 support only
• 6 & 8TB 7.2Krpm LFF HDD support
• Upto 96TB raw capacity per node
HX 3.0
36. Automated Availability Zones
Up to 32 HX
nodes
Up to 32 compute nodes
64 Node Scale with Resiliency
Cloud Scale Data Platform
64 Node & Capacity Scalability
IOVisor IOVisor
IOVisor IOVisor
IOVisor
IOVisor
IOVisor
IOVisor
Availability
Grp 3
Availability
Grp 2
Availability
Grp 1
HX 3.0