Post on 16-Jul-2015
Separating Cloud Hype from Reality in Healthcare – a
Real-Life Example at HMA
Tim Graf, VMware
Matthew Ritchart, Health Management Associates
PHC5254
#PHC5254
2
VMware Sr. Systems Engineer for Healthcare
Supports over 50 healthcare providers in Florida
17 years experience designing, building & delivering IT solutions
10 years working with virtualization technologies
Matthew Ritchart
Operations Manager for Heath Management Associates Inc.
6 years in Healthcare IT
3 years directly supporting a hospital environment
Responsible for virtual architecture and direction
About Us
Tim Graf
3
Today’s Discussion
Moving into the future of Healthcare IT
What HMA is doing that’s different…
And Why they’re doing it.
Lessons learned along the way
How to get there in a hurry
5
• Rapid consolidation of key industry players underway
• Not just hospitals, but doctor practices as well
Industry consolidation
• Bring Your Own Device mentality entering Healthcare
• Reliable, available, cost-effective cloud architecture
• Users are much more aware of technology and their access to it
Macro changes
• Changing the mindset of buying and implementing to producing and optimizing
• Provide better structured and more agile applications faster
Rapid deployment of resources
• Rapid replacement of aging healthcare systems
• Advanced technology needed to produce better results with less resources
Technology opportunity
Healthcare Moving to the Future
6
99% virtualized
Over 250 ESXi hosts running 4,000 virtual machines
1.5% IT spend
80% Meaningful Use 2.0 compliant
HMA IT
Rapidly adding facilities through acquisition
Hybrid cloud directive from management
Infrastructure as a Service
Key Stats
Strategic Objectives
7
The cloud fixes everything!
• Not the answer to all of IT management's woes
• Does not solve process or policy issues
• Simply a way to deliver applications and infrastructure
The Cloud never goes down!
• Jan 10th 2013; Dropbox is down for ~16 hours
• Jan 31st 2013; Amazon’s own website is down for 49 minutes
• February 1st 2013; Microsoft Office 365 is down for ~2 hours
Cloud storage is cheaper than physical storage!
• Is the savings worth losing control of your data?
• Even Gartner currently only suggests cloud storage as an alternative for non-mission-critical applications
• How would you recover from truly lost data? What are the consequences?
Service Level Agreements means we never have to worry about explaining outages!
• Make sure to understand what exactly is promised in the SLA
• Recuperation on monetary loss does not resolve user experiences with failed services
Common Cloud Hype
8
Software-Defined Data Center
All infrastructure is virtualized and delivered as a
service, and the control of this data center is
entirely automated by software. Abstract. Pool. Automate.
9
Extending the Private Cloud to the Hybrid Cloud
Seamless
Extension of
the Data Center
Public Cloud
Software-Defined
Data Center
Private Cloud
Any Application, Any Place… No Changes
Software-Defined
Data Center
Common Management &
Orchestration Platform
Unified Network
Architecture
Common Security Model
One Support Call
10
VMware Sr. Systems Engineer for Healthcare
Supports over 50 healthcare providers in Florida
17 years experience designing, building & delivering IT solutions
10 years working with virtualization technologies
Matthew Ritchart
Operations Manager for Heath Management Associates Inc.
6 years in Healthcare IT
3 years directly supporting a hospital environment
Responsible for virtual architecture and direction
About Us
Tim Graf
11
Agile Healthcare Platform
• Infrastructure and Operations was tasked with
building a new platform for healthcare, more advanced
than anything we had produced or seen.
• It needed to have the scalability of a public cloud while
retaining patient data on premise.
• We were also tasked with monitoring, maintaining,
automating and securing the platform with a minimal
operations time investment.
12
New goals for I&O have shifted our priorities and needs
• Just keeping the systems running is not enough
• Proper management, monitoring, documentation and capacity planning
• High speed and reactive deployments
Focus shifts from maintaining to producing
• Implementing new technology
• Designing new solutions
• Planning for future development
High Stability and availability requirements
• We run systems that affect patient’s lives
• Stability needs to match or exceed previous generations
In-house Application Development
• HMA is designing and rolling our in house developed applications
• Development requires an entirely different mindset and offering
New Mentality for HMA IT
13
• 99.999% Uptime (5.26 min/year downtime) High Availability
• Zero impact failovers
• Minimal or no manual intervention
Seamless Fault Tolerance
• Systems need to be always “up” at the infrastructure level
• Systems also always need to be “available” at the user level Zero Downtime
• Not just high performing infrastructure
• Perceived at the user level to be high performance High Performance
• Privacy, Recovery, Audit
• HIPAA, Sarbanes–Oxley Security
• Enable easy scaling as required for projects
• Predictive assessments to always have capacity available Growth
HMA Platform Goals
14
To provide all the requested features we had
to step around the usual datacenter build
and produce a physically separated
environment that was virtually local. It had
to grow as we did and still be manageable.
That’s where the vCloud Suite came in.
The Plan: Cloud-Ready Datacenter
15
vSphere/vCloud
• Scalability
• Cost
• Manageability Operations Manager
• End-to-end Visibility
• Sprawl containment
HA/FT Stretched Clusters
• Zero Downtime
Storage DRS
• Built for Growth
• High Performance
• Cost Optimization
Business Manager
• Cloud Cost tracking
Configuration Manager
• Cost Management
• Granular control
Impactful vCloud Solutions
Automation Center
• Rapid provisioning
• Self-Service
18
Not all applications provide high availability
• Single threaded sections or processes
• Non-load balanced front end
• Connections to other single threaded applications
Identify application availability failures
• Press vendors/developers on HA strategy
• Communicate dependencies or issues to project owners
Provide workarounds
• VM Fault Tolerance
• VM High Availability
• Database mirroring
All Applications Are Not Created The Same
19
Match your biggest requirements
• Project Identification
• Application role
• Location
• Production, test, development
• Backup policy
Produce a written naming policy
• Explain advantages
• Important to get buy-in
Always conform
• Everyone matches the convention
• Once one application gets by, they all will
NAMING CONVENTIONS!!!!!
20
Agile Thin Provisioned Environment
• 700VMs, 140TB
• 6 months of production
• 10 Applications
Legacy Thin Provisioned Environment
• 2,500 VMs, 350TB
• Over 5 years of production
• 100+ Applications
Storage DRS With Thin Provisioning
21
Allows standardized LUN and Datastore creation
• Pick a maximum datastore size and set standards for builds
• Drives down over-allocation of storage with pre-set sizes
Helps quickly identify storage trends and allocation
• You can view cluster storage resources as a whole
• No need for individual management of LUNs or Datastores
Datastores can move freely across the entire S-DRS pool
• Remove “Keep VMDKs Together” settings to allow free movement
• Will auto-balance I/O and Storage without storage admin intervention
Easy additions of new LUNs allows for quick scalability
• Add a LUN; make a Datastore; add to S-DRS cluster; profit
• Allows S-DRS to handle moving VMs to new Datastores
S-DRS Pools to Save Storage Management Time
22
Pool All Of The Things!
Application per Cluster Mentality
•Created a resource management nightmare
•Constant monitoring and hardware additions needed
•Drastic hardware differences
•Overly complicated hardware maintenance
•Storage must be monitored per application, per datastore
Pooled (IaaS) Mentality
•All clusters are now general clusters with the same N+1 failover
•Hardware management is one platform, scalability is limitless
•Once a cluster is “loaded” and base-lined, minor trending is required
•Applications can easily span clusters and can be live motioned
•Storage can be monitored per SDRS cluster
23
A New Standard for Agility
Storage/ Availability Servers Networking Security Management/
Monitoring
2008 2012 SDDC
Weeks Days/ Hours
Minutes/ Seconds
Software-Defined Data Center Services
Virtual Data Center
24
Cloud Path: VMware vCloud Suite
Management
VMware vCenter™
Operations
Management Suite™
VMware vFabric®
Application Director
Physical Infrastructure (Server, Storage, Network)
VMware vSphere®
Cloud Infrastructure
Software-Defined
Networking and Security
Software-Defined
Storage and Availability
Extensibility
VMware vCloud
APIs
VMware vCloud
Connector™
VMware vCenter
Orchestrator™
Virtualization
VMware® vCloud®
Automation Center
VMware vCloud Director®
VMware vCloud
Networking and
Security™
VMware vCenter
Site Recovery
Manager™
25
Choose the virtualization platform that’s proven
to deliver the lowest total cost of ownership.
Why VMware
Choose the only IT partner that’s 100% focused
on the future of your IT infrastructure.
Choose the only IT partner that gives you the
ability to virtualize, automate, and control all
data center resources.
Choose the only virtualization platform that’s
battle-tested in the largest, most demanding
enterprise environments.
Build for the
Future
Virtualize
Everything
Maximize
Uptime
Minimize TCO
27
Contact Information
Any questions, comments, or just want to chat about virtualization,
please feel free to shoot us an email
Matthew Ritchart
• MATTHEW.RITCHART@HMA.COM
Tim Graf
• TGRAF@VMWARE.COM