Stop Flying Blind: Best Practices for Optimizing your SAN and Virtualization Projects Christopher Carlton Storage Team Lead Health Technology Solutions JPS Health Network
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Stop Flying Blind: Best Practices for Optimizing
your SAN and Virtualization Projects
Christopher CarltonStorage Team Lead
Health Technology SolutionsJPS Health Network
JPS Health Network• Who is JPS Health Network?
– A leading health care service provider and only Level 1 Trauma Center, serving Tarrant County, Texas.
– A nationally recognized teaching hospital.
• Our Mission – To improve the health status of the families and
individuals in the communities we serve.
Presenter
Presentation Notes
We have multiple facilities throughout the County and they include: 567 bed county hospital including emergency room, surgical services, and psychiatric inpatient services. 30 bed Outpatient/Orthopedic Surgery Center 24 ambulatory health centers. 6 dental clinics. 3 health centers for women. 17 school-based clinics. 3 Tarrant Counties correctional facilities.
Challenges of Evolving Business
• Doing More With Less • Environment Complexity• Low Utilization of Existing Equipment• Business Requirements Increasing• Data Requirements Expanding• Limited Staffing
Applications Supported• Clinical Applications
– Patient Clinical Records– Surgical System– Critical Care Monitor Systems – Patient and Business System Interfaces
• Business Applications– A/P and Payroll– Purchasing – Materials Management – Receiving– Data Warehouse– File share – Databases
Technology Environment• SAN
– Brocade Switches– Hitachi USP-V and AMS 2XXX Storage– IBM DS4XXX Storage
• Servers– 200 Physical Application Servers– Windows/AIX/Linux– vSphere 4.1 Enterprise Plus/VMware Virtual Center
• Tools– Virtual Instruments VirtualWisdom– Brocade DCFM/SAN Health/SAN Health Pro– HiCommand Suite– Solarwinds Orion – HP SIM– Perfmon/IOstat/NMON
Lessons Learned
• Aligning with the organizations goals.• Understanding your environment
challenges.• Adopt new technologies to improve
performance and utilization of resources.
• Establish baselines and monitor them.• Embrace change. • Make a plan and take action.
Presenter
Presentation Notes
Best Practices• Tap every storage port and collect data at the
fabric layer.• Use purpose-built tools to monitor performance
and check for bottlenecks.• Use vendor best practices.• Ensure infrastructure visibility to all teams.• Build collaborative team environment.• Implement thin provisioning to increase disk
utilization at the pool level. • Use virtual server tool set to thin provision at
the host level.
Presenter
Presentation Notes
Best Practice Example: Monitor ISLs
• Proactively detect ISL physical errors before they affect users.
Best Practice Example: Measure Cache Performance
• Real-time views into caching
Best Practice Example: Measure LUN Performance
• 5 LUNS appear overloaded
Best Practice Example: Monitor vSphere Server Utilization
• Determine which servers are CPU & I/O bound
Best Practice Example: Monitor VMware Read/Write Latency
• vCenter latency data uses averages; use VirtualWisdom to measure actual read/write latency
Best Practice Example: Monitor Memory Utilization
• Indicates that there are underling memory configuration issues.
Presenter
Presentation Notes
I am curious about the level of swamping occurring on one of the ESX servers (da1vmhst201) and two of the VM’s DA1FWMON101 and DA1PMAP101. The swapping measured here by Vmware is not the host level operating system memory being swapped out but actually Vmware swapping memory in and out from underneath the host. Normally it handles memory issues more efficiently by doing what the call Ballooning. Ballooning is asking the OS if there is any memory that can be freed and using that instead because the OS believes that the memory isn’t in use and it is usually right so this memory can remain swapped out. I see for these two VM there is no ballooning going on so it looks like it is having to trade memory back and forth between two VM’s because it is memory that is actually in use. It looks like the memory utilization on these VM is actually relatively low so I am not sure why it is having to swap. ��There are configurations in Vmware that can help. One of which is reserving the memory which is recommended for certain applications like Java where the guest OS can not figure out what memory is in use so the ballooning doesn’t work.��I am hoping that you can pass this info along to the Vmware team to see if 1) they care because users have been or might be experiencing problems and 2) if they can do anything to improve it that we can then show the results and improvement. ��Also pass along my apologies to them if I clobbers some of the terminology and descriptions to them because I am still new to Vmware and this information but from what I have been learning this looks important.
Best Practice Example: Measure LUN Utilization
• Determine which file systems are most over allocated based on percentage of utilization.
Best Practice Example: Measure Storage Utilization
• Determine which has the most available space.
Summary
• Greater visibility into virtualized infrastructure.
• Rapidly identify and eliminate problems.• Increased reliability and performance.• Prepared for the challenges of server
and storage virtualization.• We are able to plan more effectively.• Our hospital administration is happy.
Questions?
Christopher CarltonStorage Team Lead
Health Technology SolutionsJPS Health Network
Related Documents
