Virtualization Myths and Realities Exposed The Circuit
1 like485 views
Bill Cashman from Peak 10 presented on virtualization myths and realities. The presentation addressed common misconceptions about the costs and capabilities of virtualization. It discussed how smaller organizations can benefit from virtualization using free hypervisors or virtual utility hosts. It also addressed myths around shared storage requirements, VMware costs, application compatibility, CPU and storage performance for virtualized databases. Real-world examples were given to show how SQL Server and Exchange can run effectively in virtual environments.
![Virtualization Myths and Realities: MYTH
“I’m looking elsewhere because VMware is the most expensive platform.”
► The “free” hypervisors don’t have Enterprise-level features.
• Hyper-V requires System Center Virtual Machine Manager.
• XenServer requires XenCenter
• VMware ESXi is free and is directly comparable to these free
editions.
► VMware does potentially have the highest initial software cost
► VMware scales more cost-effectively than Hyper-V or XenServer.1
• The recommended cost-comparison metric is cost per payload
• As an infrastructure scales, resource efficiencies of scale allow
VMware to “Do more with less.”
o Higher I/O Scalability
o Transparent Page Sharing (RAM Deduplication)
o Large memory pages
o 8-way Virtual SMP
[1] Taneja Group Technology Analysts - 2009
ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 10](https://image.slidesharecdn.com/virtualizationmythsandrealitiesexposed-100610120447-phpapp02/85/Virtualization-Myths-and-Realities-Exposed-The-Circuit-10-320.jpg)
![Virtualization Myths and Realities: MYTH
“My existing server has multiple cores, how can my app live without them?”
► Industry average server CPU utilization is between 5%-7%.1
• Industry average peak CPU utilization is 22%
for less than 30 minutes daily.
► Industry average CPU speed is 2.23GHz.1
► 7% (CPU average) of 2230MHz is 156 MHz.
• 22% (CPU average peak) of 2230 MHz is 491 MHz.
► Custom-developed apps should be fully tested prior to virtualizing production
[1] Gartner - 2010
ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 14](https://image.slidesharecdn.com/virtualizationmythsandrealitiesexposed-100610120447-phpapp02/85/Virtualization-Myths-and-Realities-Exposed-The-Circuit-14-320.jpg)
![Virtualization Myths and Realities: MYTH
“SQL and Exchange are too transactional to run effectively in virtual machines.”
SQL and Exchange are fully-supported by Microsoft as Virtual Servers
• Current version hypervisors are 95%+ CPU efficient compared to
physical hardware.1
• Modern virtual hosts are capable of sustaining 350,000+ IOPs2
• Individual VMs are capable of sustaining 120,000+ IOPs
[1] Performance Labs, Inc. – 2010 | [2] VMware, Inc - 2009
ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 16](https://image.slidesharecdn.com/virtualizationmythsandrealitiesexposed-100610120447-phpapp02/85/Virtualization-Myths-and-Realities-Exposed-The-Circuit-16-320.jpg)
![Virtualization Myths and Realities: MYTH
“SQL and Exchange are too transactional to run effectively in virtual machines.”
A vSphere hosted DB cluster holds the world records for database
performance and price/performance in an x86 architecture.
• Ran 1,316,882 Composite Queries per Hour (QphH), 13% higher than
the next best performance.
• Used 37% less server hardware than that used for the second best
performance result.
• Achieved a price/performance of US $0.70 per QphH, 7.7 times better
than that of the previous performance record holder.
• Took 16 minutes and 23 seconds to load 1 TB database at a rate of
3.7 TBs per hour, a load time 8.7 times faster than that of the
previous performance record holder.
[1] TPC-H Benchmarks, Inc. – 2010
ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 17](https://image.slidesharecdn.com/virtualizationmythsandrealitiesexposed-100610120447-phpapp02/85/Virtualization-Myths-and-Realities-Exposed-The-Circuit-17-320.jpg)
Virtualization Myths and Realities Exposed The Circuit
- 1. VIRTUALIZATION MYTHS & REALITIES PRESENTED BY: Bill Cashman, VCP Solutions Design Manager Peak 10 Data Center Solutions June 10, 2010 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA FOCUS ON YOUR BUSINESS
- 2. Agenda ► Cost ► Capability ► Compatibility ► Storage Components ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 2
- 3. Who is Peak 10? Peak 10 Customer Portal Interoperability Customer Private Cloud Traditional Data Center Services Colocation, Co-Managed, Peak 10 Enterprise Cloud Managed Hosting, Managed Services Infrastructure as a Service Plus 10 Locations/18 Data Centers | Private High-Performance Network Hosting | Hypervisor | Storage | Engineering ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 3
- 4. Peak 10 Datacenter Infrastructure High-Performance Public and Private Network Infrastructure Peak 10 utilizes multiple Tier-1 providers throughout its network of geographically diverse data centers, providing clients with the highest-level of performance available. In addition, Peak 10 maintains a robust private network, giving clients additional levels of security and service for their critical operations. Peak 10 Benefits ► Multiple Tier-1 Providers ► Redundant Network Design ► Robust Internal & External Network ► Network Monitoring and Route Optimization ► Geographic Diversity ► 24x7x365 Support ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 4
- 5. Virtualization Myths and Realities: MYTH “My business has only a few (1-3) servers. My infrastructure is too small to realize the financial benefits of Virtualization.” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 5
- 6. Virtualization Myths and Realities: MYTH “My business is too small to realize the financial benefits of Virtualization.” ► Virtual Utility Hosts • Multi-tenant infrastructure to enable many companies to share the expense and the benefit of virtualization. • Many Virtual Hosts are auditable and can meet many compliance levels. • Before investing, be sure to choose the host with the correct: o Audit & Compliance levels o Service Level Agreement ► Free Hypervisors • All major hypervisors have a free version. • Will sacrifice top-end features (live migration, HA) • Financial benefit of consolidation can be realized with as few as 2 servers. ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 6
- 7. Virtualization Myths and Realities: MYTH “The requirement for shared storage makes Virtualization an expensive proposition.” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 7
- 8. Virtualization Myths and Realities: MYTH “The requirement for shared storage prices me out of the Virtualization market.” ► Shared Storage is only required for Enterprise-level features. • Live Migration, High-Availability and clustering require shared storage. • Without shared storage you can still see benefits of Virtualization. o Server Consolidation o Streamlined Management o Rapid Deployment and Reconfiguration ► There has been a major revolution in storage driven by Virtualization. • Many storage systems exist for companies with modest budgets. o iSCSI • Equallogic • Lefthand o Fiber Channel • HP MSA • EMC AX-4/5 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 8
- 9. Virtualization Myths and Realities: MYTH “I’ve decided to look at other platforms because everything I hear says VMware is the highest-cost Virtualization platform.” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 9
- 10. Virtualization Myths and Realities: MYTH “I’m looking elsewhere because VMware is the most expensive platform.” ► The “free” hypervisors don’t have Enterprise-level features. • Hyper-V requires System Center Virtual Machine Manager. • XenServer requires XenCenter • VMware ESXi is free and is directly comparable to these free editions. ► VMware does potentially have the highest initial software cost ► VMware scales more cost-effectively than Hyper-V or XenServer.1 • The recommended cost-comparison metric is cost per payload • As an infrastructure scales, resource efficiencies of scale allow VMware to “Do more with less.” o Higher I/O Scalability o Transparent Page Sharing (RAM Deduplication) o Large memory pages o 8-way Virtual SMP [1] Taneja Group Technology Analysts - 2009 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 10
- 11. Virtualization Myths and Realities: MYTH “My code/application will break in a virtual environment because of how Virtualization changes the Operating System.” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 11
- 12. Virtualization Myths and Realities: MYTH “Virtualization will break my application.” ► All of the major modern hypervisors natively support many Operating Systems. • Hyper-V – All Microsoft 2008 editions and SUSE versions • XenServer - Windows editions from 2000 SP4 and 40+ Linux editions • VMware – Every Windows edition 3.51+ and 60+ Linux Editions ► All current-version Microsoft products are fully supported and tested on VMware, Hyper-V and Citrix XenServer ► Oracle DB has limited support on VMware, Hyper-V and Citrix XenServer • Fully-supported on OracleVM ► Custom-developed apps should be fully tested prior to virtualizing production ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 12
- 13. Virtualization Myths and Realities: MYTH “My application currently runs on a server with a dual-core 2.4GHz processor. How can I possibly replace this with a VM with only 1GHz of processor?” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 13
- 14. Virtualization Myths and Realities: MYTH “My existing server has multiple cores, how can my app live without them?” ► Industry average server CPU utilization is between 5%-7%.1 • Industry average peak CPU utilization is 22% for less than 30 minutes daily. ► Industry average CPU speed is 2.23GHz.1 ► 7% (CPU average) of 2230MHz is 156 MHz. • 22% (CPU average peak) of 2230 MHz is 491 MHz. ► Custom-developed apps should be fully tested prior to virtualizing production [1] Gartner - 2010 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 14
- 15. Virtualization Myths and Realities: MYTH “Everything I read as well as my years of experience tell me that my transactional SQL Server/Exchange Server cannot possibly run efficiently as a virtual.” ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 15
- 16. Virtualization Myths and Realities: MYTH “SQL and Exchange are too transactional to run effectively in virtual machines.” SQL and Exchange are fully-supported by Microsoft as Virtual Servers • Current version hypervisors are 95%+ CPU efficient compared to physical hardware.1 • Modern virtual hosts are capable of sustaining 350,000+ IOPs2 • Individual VMs are capable of sustaining 120,000+ IOPs [1] Performance Labs, Inc. – 2010 | [2] VMware, Inc - 2009 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 16
- 17. Virtualization Myths and Realities: MYTH “SQL and Exchange are too transactional to run effectively in virtual machines.” A vSphere hosted DB cluster holds the world records for database performance and price/performance in an x86 architecture. • Ran 1,316,882 Composite Queries per Hour (QphH), 13% higher than the next best performance. • Used 37% less server hardware than that used for the second best performance result. • Achieved a price/performance of US $0.70 per QphH, 7.7 times better than that of the previous performance record holder. • Took 16 minutes and 23 seconds to load 1 TB database at a rate of 3.7 TBs per hour, a load time 8.7 times faster than that of the previous performance record holder. [1] TPC-H Benchmarks, Inc. – 2010 ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 17
- 18. Case Closed Questions? Bill Cashman: bill.cashman@peak10.com ATLANTA | CHARLOTTE | CINCINNATI | JACKSONVILLE | LOUISVILLE | NASHVILLE | RALEIGH | RICHMOND | SOUTH FLORIDA | TAMPA WWW.PEAK10.COM 18