from virtualization 2.0 to enterprise cloud: technologies and … · 2013. 2. 26. · from...
TRANSCRIPT
From Virtualization 2.0 to Enterprise Cloud: Usages and Technologies Bing WangProduct Marketing Manager, Virtualization TechnologiesData Center Group, Intel
DCCS002
• Data Center Evolution• A Framework for Optimization• Technology Drill-down• Summary
Agenda
2
• Data Center Evolution• A Framework for Optimization• Technology Drill-down• Summary
Agenda
3
“Cloud Computing is THE critical ingredient for the next wave of IT Growth.”
Frank Gens, IDC
SVP and Chief Analyst
“Virtualization is the bridge to consuming Cloud Computing”
IDC, January 2009
4
Virtualization: Evolving Towards the Enterprise Cloud
Consolidation: Virtualization 1.0Operational Expense Efficiency
Enterprise Cloud:
Virtualization 3.0Automation and Resource Scalability
Flexible Resource Management: Virtualization 2.0
Dynamic Resource Allocation
5
Virtualization Data Center Foundation
Infrastructure Scale
Massive Scaling
Balanced Platform
RAS is critical
Unified Network(10Gb Ethernet for Storage and Networking)
Intel® Xeon® Processor Based Platform (CPU & I/O performance, New Technologies)
Lower TCO
Network
consolidation
Multi-Tenancy
Resource Sharing
Security
Isolation
Data
Cen
ter
Req
uir
em
en
tsD
ata
cen
ter
Fo
un
dati
on
• Data Center Evolution• A Framework for Optimization• Technology Drill-down• Summary
Agenda
7
8 88
Intel® Virtualization TechnologyComprehensive platform capabilities to achieve native performance, with best-in-class reliability,
security, and scalability
Intel® VT for IA-32 & Intel® 64(Intel® VT-x)Native performance of
virtualized workloads with security and reliability
Intel® VT for Directed I/O
(Intel® VT-d)Performance, reliability and security through dedication of system
resources
Intel® VT for Connectivity
(Intel® VT-c)Performance, and
scalability through a dynamically sharable
converged high-capacity interconnect
ProcessorChipset
Network
Intel® VT – Vision and Directions
8
A Framework for Optimizing Virtualization
VMM
OS
Host
OS OS…
… and scale out across VMs
Reduce overhead
within each VM…
Ways to Reduce Overheads• VT-x Latency Reductions
• Virtual Processor IDs
• Extended Page Tables
• APIC Virtualization (Flex Priority)
• I/O Assignment via DMA Remapping (VT-d)
• VMDq
Technologies for Scaling• Scaling from EP to EX
• Hyper Threading
• PAUSE-loop Exiting
• SR-IOV
• RAS Capabilities
Let’s look at the details…
VT-x = Intel® Virtualization Technology for IA-32, Intel® 64 and Intel® Architecture9
• Data Center Evolution• A Framework for Optimization• Technology Drill-down• Summary
Agenda
10
Memory
Optimizing VT Transition Latencies
• Virtual Machine Control Structure (VMCS)– VMCS holds all essential guest and host register state– “Backed” by host physical memory– Accessed via an architectural VMREAD / VMWRITE interface– Enables CPU implementations to cache VMCS state on-die
• Virtual Processor IDs (VPIDs)– VMM-specified values used to tag microarchitectural structures (TLBs)– Used to avoid TLB flushes on VT transitions
VMM
pCPU
VMREAD VMWRITE
BackingPage
inMemory
VMCS
Significant VT latency reductions over time…
0200400600800
10001200140016001800
2007 2008 2009 2010
Round-Trip VM exit/entry (Cycles)
11
Intel® Hyper-Threading Technology
• Intel® Hyper-Threading Technology (Intel®HT Technology) – Run 2 threads at the same time per core– Hide memory latencies– Higher utilization of 4-wide execution engine
• Significant benefits to virtualization– Abundant thread parallelism– Multiple vCPUs per VM– Multiple VMs per system
Net Result: Increased Scaling on Virtualization Workloads
Tim
e (p
roc.
cyc
les)
w/o HT HT
Note: Each box represents a processor
execution unit
12
Lock-holder Preemption
• A given virtual processor (vCPU) may be preempted while holding a lock
• Other vCPUs that try to acquire lock will spin for entire execution quantum
13
Solution: PAUSE-loop Exiting
• PAUSE instructions often used in spin locks– A long sequence of PAUSE’s close together signals
likely lock-holder preemption
• PAUSE-loop Exiting enables VMM to specify:– Gap: Maximum expected time between successive
executions of PAUSE in a loop– Window: Maximum time to spend in a PAUSE loop
before causing a VM exit
• Upon a PAUSE-loop Exit– VMM can then regain control and schedule another
vCPU to run
14
Example: PAUSE-loop Exiting
Executions of PAUSE in instruction stream
spin_lock:attempt lock acquire;if fail {
PAUSE;jmp spin_lock
}
Likely lock-holder preemption:
Likely normal locking behavior:
Instruction stream (time)
WindowGap
GapWindow
PAUSE-loop Exiting Performance
16
• Pause-Loop Exiting (PLE) delivers ~40%performance improvement in over-committed case
• CPU Efficiency stays flat with PLE which implies a better scaling
Server
I/O Evolution is ImminentFlexible Resource Architectures
VM VM
Converged I/O
Intel
HBA NIC
SAN LAN
Fungible and Secure I/O architecture is essential for a flexible enterprise cloud
Isola
tion
17
Infrastructure Scale: I/O Virtualization with SR-IOV
Drive adoption of SR-IOV for I/O scaling
Natively Share I/O device with Multiple VMs
Standards based
Reduces VMM overhead
Works with VM migration
Inherently provides inter-VM data isolation
VM VM VM
SR-IOV I/O device
18
Citrix* Standards-based I/O Virtualization with Intel® Xeon®
Processor 5600
Citrix and Intel Deliver Flexible and Scalable High Performance
I/O Virtualization Solution via PCI-SIG* SR-IOV and Intel VT-d
02468
101214161820
Non SR-IOV SR-IOV
2 VPXs Aggregated Throughput
Gbps
Please come to see the demo at Booth 1096
7X
Intel® VT for Directed I/O (Intel® VT-d)19
RAS at Every Part of the Platform
Socket RAS• Corrected Machine Check Interrupts (CMCI)• Recoverable Machine Check Architecture (MCA)
Memory RAS• Replay on CRC error• Memory thermal management• Memory Migration, Mirroring, and hot plug
Intel® QPI RAS• Link recovery and self-healing• Poison forwarding• Hot plug socket, hot plug IOH• Domain partitioning
IO Hub (IOH) RAS• PCI Express* Technology hot plug• PCI Express Technology Advanced
Error Reporting (AER)
Intel® QPI
Xeon 7500
Xeon 7500
Intel® Xeon®
processor 7500
Xeon 7500
PCI Express* 2.0PCI Express* 2.0
MemoryMemory
ICH10
IOH IOH
MemoryMemory
Intel® QPI = Intel® QuickPath Interconnect
20
21
HW Un-correctable Errors
Machine Check Architecture RecoveryFirst Machine Check Recovery in Intel® Xeon® processor based SystemsPreviously seen only in RISC, mainframe, and Itanium®-based systems
*Errors detected using Patrol Scrub or Explicit Write-back from cache
Allows Recovery From Otherwise Fatal System Errors
Normal StatusWith Error Prevention
System Recovery with OS
Error Corrected
Error Detected*
Error Contained
HW Correctable ErrorsUn-correctable Errors
System works in conjunction with OS or
VMM to recover or restart processes and
continue normal operation
Bad memory location flagged so data will not
be used by OS or applications
Error information passed to OS /
VMM
Making Resource Sharing Safer
VM VM VM VMIsolate
Intel® Virtualization and Intel® Trusted Execution Technology (Intel® TXT) work
together to better isolate VMs
MeasureIntel® TXT measures VMM for launch
protection
EncryptIntel® New instructions in Intel® Xeon® 5600
series quickly encrypts data in flight and at rest
Virtual Machine Monitor
Intel® TXT and New instructions in Intel® Xeon® 5600 series Make Multi-Tenancy More Secure
DCCS003: Data Center Protection Technologies
22
Intel Virtualization Performance (by processor generations)
Intel® Xeon® Processors Optimized for Virtualization
23
Inte
l Est
imat
e
Inte
l Est
imat
e
Inte
l Est
imat
e
Inte
l Est
imat
e
Unless noted otherwise, source: http://www.vmware.com/products/vmmark/results.html (Best published Intel platform scores as of 3/30/10)1. Source: http://www.cisco.com/en/US/prod/collateral/ps10265/ucsB250-VMmark.pdf2. Source: ftp://ftp.software.ibm.com/eserver/benchmarks/IBM_x3850X5_VMmark_Independent_Publication_033010.pdf
9
25
36
1421
72
2S Intel Xeon
X5470 3.33 GHz
4-core
2S Intel Xeon
X5570 2.93 GHz
4-core
2S Intel Xeon
X5680 3.33 GHz
6-core
4S Intel Xeon
X7350 2.93 GHz
4-core
4S Intel Xeon
X7460 2.66 GHz
6-core
4S Intel Xeon
X7560 2.26 GHz
8-core
VMmark* Scores
1 2
Virtualized SQL* Database Performance Large Healthcare Provider Case Study
HP* DL380 G6 w/ VMware vSphere* 4 and two VMs double native DL380 G5 transaction rate at less CPU utilization
G6 is much more power efficient at load and at idle
Response time differences are nominal from native G5 to Virtualized G6
Power in Watts
G5 G6
Avg. Test 308 272
Peak Test 372 294
Lowest Test 270 252
Avg. Idle 254 140
Peak Idle 256 142
Lowest Idle 253 137
2X Transactions, Lower CPU Utilization, Lower Power Envelope
3
Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, visit Intel Performance Benchmark Limitations -- Source: Intel Labs. See backup for detailed configurations24
Example of Direct AssignmentFedEX* Case Study
Intel 2x10GbE Directly connected
Directly Assigned to VM
File Transfer Direction
Intel® Xeon® ProcessorX5500 Series
VMware* ESX* 4.0
VM1 (8vCPU)
RHEL* 5.3
File Transfer Applications
Intel® Xeon® ProcessorX5500 Series
VMware ESX 4.0
VM1 (8vCPU)
RHEL 5.3
File Transfer Applications
Near Native Performance with
VMDirectPathutilizing Intel® VT-d
File Copy AppsApps limited by Crypto
IOV – Real Value in the Real World
Intel® Virtualization Technology (Intel® VT) for directed I/O (Intel® VT-d); NHM = Intel® microarchitecture, codename Nehalem
Source: Intel Labs, June, 2009. Performance measured using the netperf benchmark and the listed applications running on Intel Xeon X5560 @ 2.8GHz. See test configuration slides for more details. Actual performance may vary.
25
• Data Center Evolution• A Framework for Optimization• Technology Drill-down• Summary
Agenda
26
Summary
• Data Center compute models are evolving
• Evolution from Virtualization 2.0 to Enterprise Cloud drives:– Need for lower VM overheads– Increased scaling (VMs, vCPUs)– Improved reliability– Effective I/O Virtualization– Security
• Intel® Virtualization provides a rich technology portfolio to meet these needs
27
Session Presentations - PDFs
The PDF for this Session presentation is available from our IDF Content Catalog at the end of the day at:
intel.com/go/idfsessionsBJ
URL is on top of Session Agenda Pages in Pocket Guide
28
Please Fill out the Session Evaluation Form
Give the completed form to the room monitors as you
exit!
Thank you for your input, we use it to improve future Intel Developer Forum
events
29
Q&A
30
Legal Disclaimer• INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL® PRODUCTS. NO LICENSE,
EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL’S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL® PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. INTEL PRODUCTS ARE NOT INTENDED FOR USE IN MEDICAL, LIFE SAVING, OR LIFE SUSTAINING APPLICATIONS.
• Intel may make changes to specifications and product descriptions at any time, without notice.• All products, dates, and figures specified are preliminary based on current expectations, and are subject to
change without notice.• Intel, processors, chipsets, and desktop boards may contain design defects or errors known as errata, which
may cause the product to deviate from published specifications. Current characterized errata are available on request.
• Nehalem and other code names featured are used internally within Intel to identify products that are in development and not yet publicly announced for release. Customers, licensees and other third parties are not authorized by Intel to use code names in advertising, promotion or marketing of any product or services and any such use of Intel's internal code names is at the sole risk of the user
• Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance.
• Intel, Xeon, and the Intel logo are trademarks of Intel Corporation in the United States and other countries. • *Other names and brands may be claimed as the property of others.• Copyright © 2010 Intel Corporation.
31
Risk FactorsThe above statements and any others in this document that refer to plans and expectations for the first quarter, the year and the future are forward-looking statements that involve a number of risks and uncertainties. Many factors could affect Intel’s actualresults, and variances from Intel’s current expectations regarding such factors could cause actual results to differ materially from those expressed in these forward-looking statements. Intel presently considers the following to be the important factors that could cause actual results to differ materially from the corporation’s expectations. Demand could be different from Intel's expectations due to factors including changes in business and economic conditions; customer acceptance of Intel’s and competitors’ products; changes in customer order patterns including order cancellations; and changes in the level of inventory at customers. Intel operates in intensely competitive industries that are characterized by a high percentage of costs that are fixed or difficult to reduce in the short term and product demand that is highly variable and difficult to forecast. Additionally, Intel is in the process of transitioning to its next generation of products on 32nm process technology, and there could be execution issues associated with these changes, including product defects and errata along with lower than anticipated manufacturing yields. Revenue and the gross margin percentage are affected by the timing of new Intel product introductions and the demand for and market acceptance of Intel's products; actions taken by Intel's competitors, including product offerings and introductions, marketing programs and pricing pressures and Intel’s response to such actions; defects or disruptions in the supply of materialsor resources; and Intel’s ability to respond quickly to technological developments and to incorporate new features into its products. The gross margin percentage could vary significantly from expectations based on changes in revenue levels; product mix and pricing; start-up costs, including costs associated with the new 32nm process technology; variations in inventory valuation, including variations related to the timing of qualifying products for sale; excess or obsolete inventory; manufacturing yields; changes in unit costs; impairments of long-lived assets, including manufacturing, assembly/test and intangible assets; the timing and execution of the manufacturing ramp and associated costs; and capacity utilization;. Expenses, particularly certain marketing and compensation expenses, as well as restructuring and asset impairment charges, vary depending on the level of demand for Intel's products and the level of revenue and profits. The majority of our non-marketable equity investment portfolio balance is concentrated in companies in the flash memory market segment, and declines in this market segment or changes in management’s plans with respect to our investments in this market segment could result in significant impairment charges, impacting restructuring charges as well as gains/losses on equity investments and interest and other. Intel's results could be impacted by adverse economic, social, political and physical/infrastructure conditions in countries where Intel, itscustomers or its suppliers operate, including military conflict and other security risks, natural disasters, infrastructure disruptions, health concerns and fluctuations in currency exchange rates. Intel’s results could be affected by the timing of closing of acquisitions and divestitures. Intel's results could be affected by adverse effects associated with product defects and errata (deviations from published specifications), and by litigation or regulatory matters involving intellectual property, stockholder, consumer, antitrust and other issues, such as the litigation and regulatory matters described in Intel's SEC reports. An unfavorable ruling could include monetary damages or an injunction prohibiting us from manufacturing or selling one or more products, precluding particular business practices, impacting our ability to design our products, or requiring other remedies such as compulsory licensing of intellectual property. A detailed discussion of these and other risk factors that could affect Intel’s results is included in Intel’s SEC filings, including the report on Form 10-Q.
Rev. 1/14/10
32
Backup Slides
33
Memory Virtualization with Intel® VT
CPU0
VMM
I/OVirtualizationVT-x
with EPT
VM0 VMn
ExtendedPage Tables
(EPT)
EPTWalker
No VM Exits
Extended Page Tables (EPT)• Map guest physical to host address• New hardware page-table walker
Performance Benefit• Guest OS can modify its own page
tables freely• Eliminates VM exits
Memory Savings• Shadow page tables required for
each guest user process (w/o EPT)
• A single EPT supports entire VM
35
Intel® VT FlexPriority: Performance Benefits
Legacy 32-bit guests frequently access MMIO TPR during I/O intensive workloads
Eliminates most VM exits due to guest TPR access, reducing virtualization overhead and improving I/O throughput.
Intel® VT FlexPriority improves performance of 32-bit guests
w/o Intel® VT
FlexPriority
w/ Intel® VT
FlexPriority
Windows* XP SP3Ntttcp Throughput
Source: Intel. Performance measured using the Ntttcp benchmark comparing network performance of a 2 VP Windows* XP SP3 32-bit guest running with and without Intel® VT FlexPriority feature. System configuration: 2x 3.20GHz Quad-Core Intel® Xeon® processor X5386
on Stoakley platform with Intel® 5400 chipset, Intel® Pro/1000 MT Dual Port Server Adapter. Actual performance may vary.
10.7x
1
10.7
Rel
ativ
e Pe
rfor
man
ce
36
VMM
Virtual Device Emulation
Intel® VT for Directed I/ODelivering I/O Performance
VM
BufferDriver A
Application
VM
VMM
I/O Device Hardware
Device A|||||||||||||||||||
Chipset
Device A|||||||||||||||||||
BufferDriver A
Emulation Based Virtual I/O Direct Assigned I/Owith Intel VT-d
BufferDriver A
ApplicationOS OS
Intel VT-d Remap
Native Guest OS Driver
No HW Changes
Eliminates Intermediate Paths and VMM Overheads
I/O Intensive Workload Benefit Most
Layer 2 Classified Sorter
RxnRx1Rx2Rx1
RxnRx1
Idle
Tx2TxnTxn
Idle
Tx2Txn
Tx1
MAC/PHYNIC w/VMDq
Virtual Machine Device Queues (VMDq)Improved throughput by offloading data sorting to the NIC
VM1(vNIC)
VM2(vNIC)
VMn(vNIC)
Layer 2 Software SwitchVMM
LAN
Transmit Path:
• Round-robin servicing
• Ensures transmit fairness across VMs
• Prevents head-of-line blocking
Receive Path
• Data packets for different VMs get sorted at the Ethernet silicon based on MAC address/VLAN tags
• Sorted data packets get parsed to the respective VMs
• Data packets being received by respective VMs
Rx1Rx2Rx1RxnRx1Rxn
Tx1 TxnTx2Tx2 TxnTxn
VMDq enhancements
• Weighted round robin Tx
• Loopback functionality
• Multicast/broadcast support
h/w available
since 2008
Native DL380 G5 DL 380 G5 w/ESX3.5 VM Native DL380 G6
Operating System Windows 2003 32BitESX 3.5 w/Windows 2003 32Bit VM ESX 4 w/Windows 2003 32Bit VM
# CPU 4 cores only 4 vCPU only 4 cores only
Memory 12 GB (6 2GB DIMMs) 12 GB (6 2GB DIMMs)12GB (3 4GB DIMMs, 1 per channel)
Network 1GbE 1GbE 1GbE
Storage
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Software SQL 2005 32bit SQL 2005 32bit SQL 2005 32bit
Workload Humana In-house application Humana In-house application Humana In-house application
Health Care Provider Test Case Configurations (1)
DL380 G6 w/ESX3.5 DL380 G6 w/ESX4.0 VM DL380 G6 w/ 2 ESX4.0 VM’s
Operating SystemESX 3.5 w/Windows 2003 32Bit VM
ESX 3.5 w/Windows 2003 32Bit VM
vShere 4 w/Windows 2003 32bit VM
# CPU 4 vCPU only 4 vCPU only 4 vCPU only per VM
Memory12 GB per VM, 24GB per svr (6 4GB DIMMs, 1 per channel)
12 GB per VM, 24GB per svr (6 4GB DIMMs, 1 per channel)
12 GB per VM, 24GB per svr (6 4GB DIMMs, 1 per channel)
Network 1GbE 1GbE 1GbE
Storage
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Dual 4G FC –•1 100Gb Lun RAID 5? – X 10k spindles for logs•1 400Gb Lun RAID 5? – X 10k spindles for database
Software SQL 2005 32bit SQL 2005 32bit SQL 2005 32bit
Workload Humana In-house application Humana In-house application Humana in-house application38
Native Server (Customer Config) ESX3.5 VM vSphere VM
NIC Settings
•VMXNET 3.0 network drivers•Set the Net.CoalesceLowRxRateand Net.CoalesceLowTxRate on the host to 4•Set the Net.vmxnetThroughputWeight on the host to 0
FC HBA Settings•PVSCI driver•Virtual Interrupt Coalescing
•PVSCI driver•Virtual Interrupt Coalescing
BIOS Settings SMT off, AWE on, PAE on SMT off, AWE on, PAE on SMT off, AWE on, PAE on
SQL Server Settings 10GB 10GB 10GB
VM Settings Large Pages
Health Care Provider Test Case Configurations (2)
39
FedEx Case Study - Test-bed Configuration
Hardware Xeon CPU X5560 @ 2.8 GHz (8 cores, 16 threads); SMT, NUMA, VT-x, VT-d, EIST, Turbo Enabled (default in BIOS); 24GB Memory; Intel 10GbE CX4 Server Adapter with VMDq
Test Methodology Ramdisk used, not disk drives. We are focused on network I/O, not disk I/O
What is being transferred?
Directory structure, part of Linux repository: ~8G total, ~5000 files, variable file size, average file size ~1.6MB
Data Collection Tool ESXTOP used to capture CPU utilizationReceive throughput captured with sar in VM
VM configuration 8 VMs and each VM is configured with 1 vCPU, 2GB RAM & RHEL * 5.3 (64 bit)
Application Tools used in VM
Netperf (common network micro-benchmark); OpenSSH, OpenSSL (standard Linux layers); HPN-SSH (optimized version of OpenSSH); scp, rsync (standard Linux file transfer utilities); bbcp (“bit-torrent-like” file transfer utility); open-iscsi(initiator); iet (iSCSI target)
Source Server
Intel Oplin 10GbE CX4
Directly connected back-to-back
Directly assigned to VM
File Transfer Direction
Destination Server
Intel Xeon X5500 Series
VMware* ESX* 4.0
VM1 (8vCPU)
RHEL* 5.3
File Transfer Applications
Intel Xeon X5500 Series
VMware* ESX* 4.0
VM1 (8vCPU)
RHEL 5.3
File Transfer Applications
40