Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
SCPHCA SAN Presentation
Transcript of SCPHCA SAN Presentation
No Single Points of Failure
Single Points of Failure
Active\Active or Active\Passive
Dual Storage Processors
Protocol & Fabric
Enterprise Storage Design Considerations
Andy Patel, Stephen Balogh | HillSouth
RAID design by Workload
Protocol & Fabric
Standby: Identical\Low Performance?
How will IOPs be Distributed
VM to Datastore Distribution
Single VMFS per LUN
Lowest Host Overhead, SCSI data encapsulated into FC Frames
: Requires 10Gb-e, Requires Jumbo Frames, no IP encapsulation = lower overhead
: flexible, can cause additional host CPU overhead due to IP encapsulation, 1Gb or 10Gb-e
: don't do it
large IO equals low IOPs and high throughput
small IO equals high IOPs and low throughput
Physical\Virtual RDM's: Pros vs Cons
VMFS per LUN
Stick with single VMFS per LUN to avoid metadata locks.
Distribute and "Mix" High IO workloads with Low IO
Logical DataStore Seperation: Masking, Zoning, etc
Hardware Disk Encryption
Direct SAN Integration
Plan for Growth
Snaps, Swap, Logging
Thin\Thick - Pros & Cons
RAID by workload: dedicated\shared\RPM\Spindle Count
UPS: Stop IO and have Time to Commit Writes
Replication Bandwidth Requirements: On\Off Site, Sync vs. aSync, Array\Off Array Based, RPO & RTO
Advanced Technology: Caching, Auto\Sub-LUN Tiering, Storage IO Control: SIOC, vStorage API's for Array Integration: VAAI
Redundancy: Parity, HotSpares
Spindle Count: IOPs, Size, licensing impact
Write Heavy: RAID 6 calculates
parity twice leading to a write
LUNs per Group\Workload per group
5 vs 6 vs 10; there's no place for 0.
Set Appropriate Queue Depth
Misc. Considerations: Queue Depth, BUS, IOPS
Misc. Considerations Contd.
Misc. Considerations: RAID.
Consider Impact on Consolidation Ratios
Too High can have a negative performance impact
Ensure all host and storage node HBA are in the same lane PCI slots
Sustained vs Burst requirements
Caching vs Actual
Misc. Considerations: Compliance
PCI, DIACAP, HIPPA/HHS, FISMA/NIST, SOC/COBIT
Physically Separate Compute Clusters
Dedicated Management Subnet
Logical Datastore Separation: masking\zoning
Hardware Disk Encryption
Complete Role Based Access Control
CTO | HillSouth
DR Connectivity & Automation
DNS - Internal
IP - External
Direct Guest\VM Integrated\SAN Integrated
VM Image\File Level
Datacenter in a Box Solutions
Focus shifting from performance hardware to Software derived benefits
Commodity x86 Hardware gaining substantial market share in enterprise storage
SMB Friendly Price Points for enterprise features as these features become software driven and "baked in":
Backups with Direct SAN Integration
VMware Replication vs Array Based
Flexible RTO and RPO Planning
Lower HW Maintenance...
Many of these features didn't exist a few years ago
Storage was about expensive Hardware requirements to meet demanding workloads
A few years ago key factors in the appeal to the small and mid-size organization was Resource Utilization\Consolidation Ratios, Redundancy, Replication and Deployment Time
It was still a substantial investment
Storage is infinitely more complex with attractive new features which dramatically reduce capital and operational expense for a scaled up, high Performance Infrastructure.
These same complexities and features often distract and deter the small environment (20-50 Server..ish).
At the leading edge of the Virtualization conversation today is All SSD Arrays, VDI, Multi-Tenancy, Self Service, Software Defined Networking, Unified Workspaces...
It's all sexy, and as many a men do I often find myself chasing complex and attractive features, even if I can't have them....
For Most Small and Medium Environments, if we stick to the original conversation about storage that performs and virtualization for consolidation, redundancy, replication and agility; a virtual environment with solid shared storage costs less than a classic physical deployment with countless additional benefits.
Can it sprawl out of control? Absolutely, if you chase complex and attractive. It can also be controlled if you stick to critical needs and actively manage capacity.
Advanced Infrastructure Engineer | HillSouth