This session will cover the pro's and con's of virtualization as well as lessons learned from real world virtualization of DCS environments. Chris has deployed virtualization in ICS with and without ICS vendor cooperation.
2.
Increased redundancy
Decreased recovery time following a failure
Hardware refresh is simpler
System upgrades / rollbacks are easier
OS / system patching are simpler – allows for a
“golden image” that can be easily patched
Deployment of additional servers/services is quicker
2
3. Virtualization for Industrial Control Systems
Can the same benefits that traditional IT
receives be realized?
The simple answer? It Depends…
3
4. When dealing with ICS and virtualization, there
are a few questions that need to be asked:
Will the vendor support it?
Are we ready culturally?
Is it technically feasible?
Is it economically feasible?
4
5.
Challenges for Adoption:
◦ Vendor Support
Proprietary hardware?
Legacy technology constraints?
◦ Cultural – IT / Control System Staff / Plant Management
Virtualization not often fully understood
Can be seen as “all eggs in one basket”
Training – sufficient skills exist?
Support – ICS Staff, MIS or a combination?
Fear of the unknown or IT takeover…
5
6.
Challenges for Adoption:
◦ Technical Feasibility
Some vendors still use proprietary hardware
Can be internal server cards or external
communication/support devices:
Fieldbus cards (Modbus, MB+, Profibus, etc.)
Ethernet devices
Other devices/restrictions?
◦ Economic Feasibility
Initial deployment costs can be high
Hidden costs?
Training
Network infrastructure
Costs typically overridden by advantages gained
6
7. Assuming we’ve made it past the first 4
questions, what does ICS virtualization look
like?
7
8.
Design Considerations
◦ Virtual Infrastructure
Recommend clusters with common storage pool
Recommend 2 clusters in separate locations
Eliminates “all eggs in one basket”
◦ Plant LAN / Process Control Network
Redundancy is the primary consideration – work to eliminate
daisy-chaining and other topology issues
Existing networks may be restricted to 100Mb/s or less virtualization requires at least 1Gb/s – preferable 10Gb/s to
avoid storage or other bottlenecks
Often times, plant network upgrades and virtualization go
hand-in-hand
8
9.
ICS Virtualization – An Approach:
◦ Select an IT industry standard platform, ex.
Cisco/NetApp Flexpod using VMware
◦ Develop virtualization standards specific to ICS
Hardware
Software
Testing/Deployment strategy
Administration
Maintenance / Life-cycle Management
9
11.
The deployment scenario:
◦ Provides for full redundancy, above and beyond
clustering within each individual environment
◦ Allows ICS redundancy to be split:
Between plant and secondary location if
desired/needed
Primary servers in plant and secondary servers in 2nd
location
◦ Highly Scalable - Allows for easy expansion
11
12.
Implementation Challenges/Caveats:
◦ Deployment:
If possible – stand up virtual infrastructure in parallel
to existing system – allow sufficient time and testing
prior to cutover
Ensure redundancy is fully tested/verified – within
virtual infrastructure and network
Look for ICS specific catches:
12
13. Example ICS Caveat
VMware DRS Cluster
• Single Cluster
• Primary & Backup HMI
Servers On Same
Host
• Host Failure
• Both Servers Down
• Operations Blinded
Host 1
BU
Host 2
PRI
Host 3
Host 4
13
14. • Single Cluster
• Cluster Divided into
Groups
VMware DRS Cluster
Host 1
PRI
“Primary” DRS Group
• Host Failure
• Backup HMI Server
Still Available
Host 2
BU
Host 3
“Backup” DRS Group
• Primary HMI Server
Moves To New Host
• Operations Is OK
Host 4
14
15. ◦ Cutover:
Proper planning is the key! A staged approach is
best…
Be prepared as any issues, related or not, will be
pinned to the virtual infrastructure…
15