how to build an enterprise archive international muse 2015 education session 1105 may 28, 2015 tim...
TRANSCRIPT
How to Build an Enterprise Archive
International MUSE 2015 Education Session 1105May 28, 2015
Tim Kaschinske, Bridgehead SoftwareJim Fitzgerald, Park Place International
Agenda
• The Data Management Challenge• Archive and Backup: A Symbiont
Circle• Extracting Value from Archived Data• Cloud Integration• Open Discussion
The Continuing Impact of Virtualization(aka the Software-Defined Data Center or SDDC)We used to assign discrete physicalResources to IT tasks:
Now we allocate those resources froma virtualized “pool”:
vDesktops
200 Laptops
2096 NetworkPorts
10 Switches
Virtualization Layer
CPU’s
Flash DrivesMechanical Drives
VM
BandwidthCores
RAMStorage Controllers
Network Controllers
2 Routers
VLAN VSAN
Windows Boot SANProductionSAN
Backup Device
vServers Virtual Ports Virtual Disks
The View from the Data CenterMEDITECH
ClusterAdministrative & General Cluster
Imaging Cluster
VM Backup
VTL/CIFSObject/BLOB Store
NASProduction Storage(Hybrid)
Production Storage
(All Flash)
SQL Backup
App Backup
File Archive BrokerImage
Archive Broker
What Did You See in The Prior Diagram ?
IT Person• Complexity• Incomplete
replicas• Many Points of
Mgmt• Duplicate IO’s• Failed Backups• Long Nights• Scale Issues• Cloud Offload
Administrator• Complexity• Risk• Compliance
Issues• Potential
Impact of Data Loss
• Expense
Clinician• Complexity• Which
application gets priority?
• Wait times• Threat to
Patient Record
Why we……Backup• Risk of Data Loss• Operational
Recovery• Disaster Recovery• Development or
Test Copies of Live System
• HIPAA, HITECH, & ARRA Compliance
Archive• To Granularly Protect Critical
Production Data• To Protect Recovery Times• To provide “Single File
Restores”• Legal Records• Compliance: SOX, Local Data
Retention Rules• To retire aging applications
without losing reference points
• Data Mining• To reduce the cost of storage
How we…...Backup• As blocks, files, and objects• With or without embedded
metadata• To Tape, Virtual Tape, Block
Stores, File Stores, Object Stores, and “The Cloud”
• Over SAN’s, LAN’s, WAN’s, VPN’s, and The Internet
• We make a copy or a “saveset”, and keep the original
• With or without encryption and deduplication
Archive• As files and objects• With metadata tracked
• To File and Object Stores and “The Cloud”
• Over LAN’s, WAN’s, VPN’s, and The Internet
• We make one or more copies and delete the original
• With or without encryption and deduplication
The Symbiont Circle
The GungansThe NabooData Backup
(Rapid Recovery of Big Things)Data Archiving
Quick Grab of Single Object
• More Archives = Smaller Backups• Faster Backups = Better RTO/RPO
• Not Independent >>> Interdependent
Technical Trends Impacting BURA• Zero tolerance for Downtime• Trend to High Availability vs Disaster Recovery
– Local & Metro Storage Clusters– VMware HA– MS-Clusters
• “Fast RTO/Low-Zero RPO” Restores from Replicas– Mirrors– Snaps– CDP and CRR “Bookmarks”– Impact of Flash
• In-Memory Backups and Restores or Simple Mirrors
• Inexpensive Private Cloud, Managed Cloud, and Public Cloud “Archive as a Service”
IT Cloud Co-generation Circa 201x
Internal Private External or Public
Applications Execution
Primary Storage
Archive Storage
Backup and Recovery
Data Repository/Analytics
Client Provisioning
Client Management
Security
Current State
Future State
So what’s the problem?• Proprietary archiving techniques often
embedded in HCIS, PACS, and email applications
• Metadata can be unwieldy• Need for “stateful backup” of specific
systems• Complexity of backing up rapidly scaling
healthcare and administrative databases• Too many “storage dumps”• Avoidance of “roach motels”• Management of compliance generations
And here’s Tim with some good ideas for solving the
problem…
Archive vs. Backup
Archive Backup
File Level Granularity System Level granularity
Full Search capability Limited search capability
Long term storage (days -> infinity)
Short to medium (days -> 5 years)
Random access Serial access
Software encryption Hardware encryption
Good for retrieving Files OK for retrieving Files
Bad for recovering systems Good for recovering systems
Application Components
Application Source
Application Database Structured Data
Unstructured Data
Protecting Application Components
Application Source
Application Database
Backup
Archive
Archive for Protection
Application Source
Application Database
Content Archiving…
…Where you know information about the patient and the procedure related to the file being archived
- Patient ID
- Patient Name
- Date of Birth
- Procedure Code
- Document Type
How to obtain content?
Completing the Picture
Bringing it all together1. If Backup and Archive are both broken, or chaotic, assess risk to
determine what to address first.– Resolving an archive strategy first, when possible, clarifies backup and
recovery strategies– Do not confuse High Availability with Disaster Recovery
2. Develop overarching designs that:– Minimize redundant hardware investments– Simplify and automate operations– Allow for easy data lifecycle migration– Are easily audited/prove compliance– Leverage cloud storage for compliance and disaster protection of archives
and/or backups– Integrate future technology capabilities (generally via standards)
3. Consider the long picture– How many generations will an archived object live?– How many clinical requests will be made?– How many legal requests will be made?– Will the data management strategy outlive the current IT strategy?
GROUP DISCUSSION
Thank you for participating !