Performance Management

for 1000 virtual servers

(SHARE SESSION 10335)

Barton Robinson
Velocity Software
[email protected]

Copyright 2011 Velocity Software, Inc. All Rights Reserved. Other products and company names mentioned herein may be trademarks of their respective
 Performance Management for 1000 virtual servers

Objectives
• Define service management
• How to define service targets
• Drill down requirements
• VM Subsystems
• Linux Subsystems
• Alerts Discussions

• OR, avoid problems, spend an extra “x”

million “cash units” to avoid problems
2

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Velocity Software Performance Management
• Instrumentation Requirements
• Performance Analysis
• Operational Alerts
• Capacity Planning
• Accounting/Charge back

• Correct data
• (rmfpms is being withdrawn, SOD)
• Capture ratios
• Instrumentation can NOT be the performance
problem

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis Skills Requirement

Performance skills pay for themselves

• Buy 16 gb of storage, or “set mdc”?

• Buy more IFLs or more DASD Cache?

• Buy more CECs or reduce storage requirements

• What solution does a hardware vendor offer, tuning or….

Free performance class? (Mountain View, CA)

ZTUNE anyone? (performance service offering)
4

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis for 1000 virtual servers

Today’s requirements:
• 1,000 virtual servers under z/VM?
• 10,000 virtual servers under ESX
• 5,000 P-series
Performance analysis requirements
• Hypervisor layer (LPAR, ESX)
• Server layer
• Application layer
• (cache one call, reduce overall response time by 50%)
• Full data capture (crystal balls not needed)
• Useable granularity

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Full data capture to diagnose problems

•To resolve, requires:

•Linux Process data before and after across ALL servers
•Virtual machine data before and after
•Disk performance data before and after
Report: ESALPARS Logical Partition Summary
Monitor initialized: 02/06/07 at 13:00:00 on 2094 serial 2BFBD
------------------------------------------------------------------
<--Complex--> <-------Logical Partition---> <-Assigned
Phys Dispatch Virt <%Assigned> <---LPAR-->
Time CPUs Slice Name Nbr CPUs Total Ovhd Weight Pct
-------- ---- -------- -------- --- ---- ----- ---- ------ ----
14:17:00 13 Dynamic Totals: 0 30 791.3 11.1 100 100
14:18:00 13 Dynamic Totals: 0 30 782.6 9.6 100 100
14:19:00 13 Dynamic Totals: 0 30 857.4 11.4 100 100
14:20:00 13 Dynamic Totals: 0 30 866.0 10.7 100 100
14:21:00 13 Dynamic Totals: 0 30 866.3 10.8 100 100
14:22:00 13 Dynamic Totals: 0 30 811.2 10.9 100 100
14:23:00 13 Dynamic Totals: 0 30 809.6 10.7 100 100
14:24:00 13 Dynamic Totals: 0 30 876.3 11.4 100 100
14:25:00 13 Dynamic Totals: 0 30 834.5 10.7 100 100
14:26:00 13 Dynamic Totals: 0 30 1187 8.8 100 100
14:27:00 13 Dynamic Totals: 0 30 1286 5.5 100 100
14:28:00 13 Dynamic Totals: 0 30 1282 5.5 100 100
14:29:00 13 Dynamic Totals: 0 30 1293 3.5 100 100

3
6
/
 Performance Analysis for 1000 virtual servers

Reality check
• 100-150 servers per z/VM LPAR
• ESX, MS, P unlimited

• System level problems solved at a system

level

• “VM SYSPLEX” Announced:

• Different set of challenges

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis for 1000 virtual servers

Analyzing 1,000 servers

• Concepts are not new
• Traditional sites ran 5,000 concurrent logged
users (virtual machines)
• Requirements for analysis are
• “the same but different”

Operational alerts required to alert

From z/VM, from Linux

Performance analysis high level granularity,

• Zoom to finer granularity
8

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Service Management

Performance Management implies Service

Contracts, Chargeback, and SLAs

Service contracts based on

• Keeping users happy
• Financial obligations

Requirements
• Detect service issues
• Resolve service issues - QUICKLY

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Defining the Service

Traditional CMS (one process per virtual server)

• CMS Response times
• Percentiles

Linux Response time

• Applications require instrumentation (ARM)
• Cost of ARM currently prohibitive (opportunity)

Linux Load measurements more common

• By application, server

10

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Detecting Service Issues

Detecting service issues requires

• Defining your workload in manageable units
• Defined normal operational parameters
• Record history of application, servers

Define “application” or “server” norms by

• CPU trends
• Storage requirements
• I/O requirements
• Network activity (bandwidth, connections)

11

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis Platform Requirements

Platforms have specific requirements

z/VM Subsystems
• LPAR Analysis, Processor
• Storage / Paging
• DASD
• Scheduler / Dispatcher
Linux Servers
• Storage / Swap
• Process Table
Other virtualization platforms have different
requirements
12

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis Large System Challenges

DASD Subsystem Analysis

• Analyze performance problems for 5,000 disks?
• Some installations have 20,000+ (20 years ago)
• Now we have “Shared DASD” that needs to be measured
Virtual Machines
• Analyze performance for 1,000 virtual machines
• 5,000 was common 20 years ago
• (150 per lpar is a reasonable maximum today)
Linux Servers
• Analyze performance for 1,000 nodes
• How many on one “LPAR” or “ESX”?
• How many LPARs or ESX Servers?
13

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 1000 User Analysis

Large system? Concepts?

1,119 users logged on, 180 active
Screen: ESAMAIN Velocity Software - VSIVM4 ESAMON 3.
1 of 3 System Overview LIMIT 500

<---Users----> Transact. <Processor> Cap-

<-avg number-> per Avg. Utilization ture
Time On Actv In Q Sec. Time CPUs Total Virt. Ratio
*------- ---- ---- ---- ---- ---- ---- *---- ----- -----
10:01:00 1119 183 13.0 29.6 0.13 1 31.4 28.0 100
10:00:00 1119 179 19.0 30.8 0.11 1 15.0 13.8 100
09:59:00 1119 181 18.0 30.1 0.10 1 15.0 13.7 100
09:58:00 1120 179 15.0 29.9 0.11 1 14.9 13.6 100
09:57:00 1120 149 19.0 29.3 0.11 1 15.3 14.0 100
09:56:00 1119 120 19.0 28.8 0.13 1 61.0 59.7 100
09:55:00 1119 123 20.0 28.3 0.11 1 14.9 13.7 100
14

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 VM “granularity”

1000 servers require grouping

Traditional “O/V”
• User (virtual machine) classification
• by function,
• users by department
Linux servers (Grouping requires working
knowledge)
• Group by application (http, printing, database)
• Department
• Application subsystem
15

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Virtual Machine “granularity” Requirements

Group virtual machines on one LPAR

Group “nodes”
• By application
• By ESX

Requirements to group applications across

nodes
• Understand loads when workload balanced across many
servers

Requirements to monitor ESX, BX, Blades

16

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 VM “granularity”

Analyze 1000 servers with classification “USERCLASS”

Screen: ESAUSP2 Velocity Software ESAMON 3.808 08/05 09:45-09:46
1 of 3 User Percent Utilization CLASS * 2096 44B42
<-------Main Storage-------->
UserID <Processor> <Resident-> Lock <-WSSize-->
Time /Class Total Virt Total Actv -ed Total Actv
-------- -------- ----- ----- ----- ----- ----- ----- -----
09:46:00 System: 65.82 64.90 604K 558K 6407 611K 559K
SUSE 48.22 48.11 150K 150K 1475 148K 148K
TEST 5.70 5.52 190K 180K 2102 188K 178K
*TheUsrs 4.81 4.76 23636 23627 528 24671 23084
REDHAT 3.04 3.00 176K 176K 877 182K 182K
Velocity 2.99 2.90 7235 6861 21 7855 6834
KeyUser 0.66 0.29 3486 3486 1332 2239 2154
Servers 0.26 0.20 1786 1026 6 1947 1022
ORACLE 0.12 0.12 15962 15962 66 15875 15875
DEMOTH 0.01 0.00 37409 2255 0 41168 2255

17

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Virtual Machine “granularity”

Zoom on “CLASS SUSE” shows what is important

Screen: ESAUSP2 Velocity Software ESAMON 3.808 08/05 09:45-09:46
1 of 3 User Percent Utilization CLASS SUSE USER * 2096 44B42
<-------Main Storage-------->
UserID <Processor> <Resident-> Lock <-WSSize-->
Time /Class Total Virt Total Actv -ed Total Actv
-------- -------- ----- ----- ----- ----- ----- ----- -----
09:46:00 SUSELNX2 45.51 45.47 4198 4198 0 4080 4080
SLES11X 0.20 0.20 11809 11809 214 11595 11595
SLES9X 0.17 0.16 12182 12182 35 12147 12147
SLES9 0.15 0.15 13012 13012 208 12804 12804
SLES8X 0.10 0.09 35221 35221 250 34950 34950
SLES8 0.07 0.04 28332 28332 0 33975 33975
SLES10 0.05 0.05 23646 23646 219 23427 23427
SLES10X4 0.04 0.04 44438 44438 229 44209 44209
SUSEAPPS 0.02 0.02 11503 11503 11 11492 11492

User classification is worth the investment

18

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Large Disk Subsystem Analysis

DASD Subsystems – analysis for 4,000 3390s?

• Disk configuration shows 4,000 devices, granularity issue
ESADSD1 LISTING A1 V 129 Trunc=129 Size=9344 Line=0 Col=1 Alt=0
====>
00000 * * * Top of File * * *
00001 1Report: ESADSD1 DASD Configuration
00002 Monitor initialized: 11/06/10 at 16:07:10 on 2097 serial 374E2
00004 --------------------------------------------------------------
00005 Dev Sys Device <CHPIDS OnLn><-Cntrl Unit-> UserID
00006 No. ID Serial Type SHR 01 02 03 04 OBR/CU Model (if ded
00007 ---- ---- ------ ------ --- -- -- -- -- ------ ------- -------
00008
00009 0400 0000 LT35C2 3390-9 YES B4 D9 BA BB 3C/00 2107
00010 D4 D5 E6 DB
00011 0401 0001 LT35C3 3390-9 YES B4 D9 BA BB 3C/00 2107
00012 D4 D5 E6 DB
00013 0402 0002 LT35C4 3390-9 YES B4 D9 BA BB 3C/00 2107
00014 D4 D5 E6 DB
00015 0403 0003 LT35F6 3390-9 YES B4 D9 BA BB 3C/00 2107
00016 D4 D5 E6 DB
00017 0404 0004 LT35F7 3390-9 YES B4 D9 BA BB 3C/00 2107
00018 D4 D5 E6 DB 19

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Disk Subsystem Analysis

DASD Subsystems – analysis for 4,000 devices?

• What is relevant for performance? (top 10 busy devices)
Report: ESADSD2 DASD Performance Analysis Linux Te
---------------------------------------------------------------------------
<--------DASD Response
Dev Device <--SSCH--> <%DevBusy> <SSCH/sec-> <--Service time
No. Serial Type Total ERP Avg Peak avg peak Resp Serv Pend Disc
---- ------ ------ ------ --- ---- ---- ----- ----- ----- ---- ---- ----
16:09:07
***Top DASD by Device busy***
BC15 LP394B 3390-9 40334 0 44.0 44.0 672.2 672.2 0.7 0.7 0.3 0.0
EC6D LP366D 3390-9 12495 0 36.6 36.6 208.3 208.3 1.8 1.8 0.2 0.1
EF05 LP3505 3390-9 2741 0 26.3 26.3 45.7 45.7 5.7 5.7 0.3 3.2
B908 LP3B84 3390-9 12988 0 26.2 26.2 216.5 216.5 1.2 1.2 0.3 0.5
EC4A LP3641 3390-9 8134 0 26.0 26.0 135.6 135.6 1.9 1.9 0.2 1.1
EC2A LP356C 3390-9 18559 0 22.6 22.6 309.3 309.3 0.7 0.7 0.2 0.0
EC29 LP356B 3390-9 15997 0 19.2 19.2 266.6 266.6 0.7 0.7 0.2 0.0
B905 LP3B81 3390-9 9595 0 18.6 18.6 159.9 159.9 1.2 1.2 0.2 0.4
ED12 LP3547 3390-9 1643 0 15.1 15.1 27.4 27.4 5.5 5.5 0.3 2.0
EC49 LP3640 3390-9 4067 0 14.0 14.0 67.8 67.8 2.1 2.1 0.2 1.2
***End Top DASD by Device busy*** 20

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Disk Subsystem Analysis

DASD Subsystems – analysis for 4,000 devices?

• Show control units (256 devices / ctl unit?)
Report: ESADSD2 DASD Performance Analysis Linux Te
---------------------------------------------------------------------------
<--------DASD Response
Dev Device <--SSCH--> <%DevBusy> <SSCH/sec-> <--Service time
No. Serial Type Total ERP Avg Peak avg peak Resp Serv Pend Disc
---- ------ ------ ------ --- ---- ---- ----- ----- ----- ---- ---- ----
16:09:07
B800 Control Unit 1881 0 0.0 0.0 31.3 31.3 2.2 2.2 0.2 1.0
B900 Control Unit 28808 0 0.3 0.3 480.1 480.1 1.3 1.3 0.3 0.5
BC00 Control Unit 42348 0 0.3 0.3 705.8 705.8 0.9 0.9 0.3 0.2
BD00 Control Unit 3350 0 0.1 0.1 55.8 55.3 4.3 4.3 0.5 1.7
BF00 Control Unit 1451 0 0.0 0.0 24.2 24.2 4.7 4.1 0.3 2.7
EC00 Control Unit 87867 0 0.9 0.9 1465 1456 1.3 1.2 0.1 0.4
ED00 Control Unit 20413 0 0.9 0.9 340.3 336.6 5.2 5.2 0.7 1.4
EE00 Control Unit 5506 0 0.2 0.2 91.8 91.0 4.9 4.9 0.6 0.9
EF00 Control Unit 4975 0 0.3 0.3 82.9 82.9 4.3 4.1 0.3 2.3
------ --- ---- ---- ----- ----- ----- ---- ---- ----
System: 203294 0 0.1 0.1 3389 3343 1.8 1.8 690 62K
21

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Storage Subsystem Analysis

Storage Subsystem – analysis for 40 Million pages

Report: ESASTR1 4.1.0 09/08/11 Page 1
Monitor initialized: 102097 First record analyzed: 10/09/09 09:22:00
------------------------------------------------------------------------
Users <-------Pages----------------------------------------------------> Over
Loggd System <Available> Systm User NSS/DCSS <-AddSpace> VDISK <MDC> Diag Commit Capt-
Time On Storage <2gb >2gb ExSpc Resdnt Resident Systm User Rsdnt Rsdnt 98 Ratio Ratio
-------- ----- ------- ----- ----- ----- ------ -------- ----- ----- ----- ----- ---- ----- -----
09:23:00 211 40108K 284 5822 11492 38963K 6169 713K 0 2758 12 2769 2.310 0.998
09:24:01 211 40108K 212 3866 11493 38966K 5369 713K 0 2636 12 2769 2.310 0.998
09:25:00 211 40108K 131 3694 11494 38968K 4608 713K 0 1527 12 2769 2.310 0.998

Where to analyze?
• Users consuming 39M out of 40M pages

22

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Storage Subsystem Analysis

Storage Subsystem – analysis for 39 Million pages

• Which class of “Z/VM” virtual machines?
Report: ESAUSPG User Storage Analysis
-------------------------------------------
<---Storage occupancy in pages--->
UserID <---Main Storage---> <--Paging--->
/Class Total >2gb <2GB Xstor DASD
-------- ------ ------ ------ ------ ------
09:27:00 38966K 578615 38387K 2593K 64888K
***User Class Analysis***
Servers 2 2 0 858 7501
ZVPS 2950 2946 4 2031 5367
Linux 100 96 3 86 389
SOUTHLK 11685K 0 11685K 442949 14670K
APPS 140269 0 140269 16060 342250
NORTHLK 5841K 0 5841K 253513 7802K
WESTCST 3410K 0 3410K 88959 4962K
TheUsers 17883K 571131 17312K 1787K 37057K
23

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Linux Node Processor Analysis

Nodes are classified into “node groups”

• Which “NODE group” is using CPU?
• Which “esx group” is using CPU?
Report: ESAUCD4 LINUX UCD System Statistics Report
Monitor initialized: 09/10/09 at 09:21:50 on 2097
--------------------------------------------------------------------
Node/ <---------Rates Per Second------>
Date/ <Processor Pct Util> NICE <-Swaps-> <-Blocks> Switch Intrpt
Time Total Syst User Idle Time In Out In Out Rate Rate
-------- ----- ---- ---- ---- ---- ---- ---- ---- ---- ------ ------
09:27:00
***Node Groups***
SOUTHLK 36.0 25.0 11.0 2723 0 0 0 0 0 9644.8 3323.6
APPS 1.0 1.0 0 98.0 0 0 0 0 0 180.0 64.9
NORTHLK 16.0 15.0 1.0 1955 0 0 0 0 0 5539.7 1599.3
WESTCST 7.0 7.0 0 1076 0 0 0 0 0 2753.0 776.9
TheUsers 104.0 91.0 13.0 9768 0 0 0 0 0 20171 14855
24

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Node Process “granularity” Requirements

Report: ESAHSTA LINUX HOST Application Report

Monitor initialized: 09/10/09 at 09:21:50 on 2097 serial
----------------------------------------------------------------------
Node/ Process/ <Application Status Counts> <-----Processor---->
Date Application Run- Res Load <---Utilization---->
Time name Total Actv ning Wait -ed Percent seconds Avg
-------- ----------- ----- ---- ---- ---- ---- ------- ------- ----
09:23:00
***Node Groups***
SOUTHLK *Totals* 3136 270 28.0 3103 0 83.8 50.5 0.0
java 919.0 137 0 914 0 61.3 36.9 0.1
khelper 5.0 5.0 0 5.0 0 0.6 0.3 0.1
kjournal 142.0 73.0 0 142 0 12.9 7.8 0.1
snmpd 28.0 28.0 28.0 0 0 3.3 2.0 0.1
APPS *Totals* 58.0 8.0 1.0 57.0 0 1.2 0.7 0.0
java 3.0 3.0 0 3.0 0 0.7 0.4 0.2
NORTHLK *Totals* 1411 155 20.0 1391 0 35.2 21.1 0.0
java 97.0 70.0 0 97.0 0 21.9 13.2 0.2
kjournal 100.0 52.0 0 100 0 6.6 4.0 0.1
multipat 5.0 5.0 0 5.0 0 5.2 3.1 1.0
snmpd 18.0 18.0 18.0 0 0 1.0 0.6 0.1
WESTCST *Totals* 897.0 67.0 11.0 886 0 19.1 11.4 0.0
java 202.0 26.0 0 202 0 12.1 7.2 0.1
kjournal 40.0 23.0 0 40.0 0 3.5 2.1 0.1
multipat 2.0 2.0 0 2.0 0 1.2 0.7 0.6
snmpd 11.0 11.0 11.0 0 0 1.4 0.9 0.1
25

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 A scalable z/VM Performance Monitor
Traditional model (1989) VM
ESAMON: Real time analysis CP Monitor

 Uses Standard CP Monitor

Real Time Analysis
ESAMAP: Performance Reporting zMON
Post Processing Real-Time
Displays
Creates Long Term PDB
PDB or monwrite data input
PDB
PDB (Performance DataBase)
Complete data zMAP
Reports
By Minute, hour, day
PDB
Monthly/Yearly Archive
26

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Linux and Network Data Acquisition
zBX
TCPIP
WinNT SNMP/MIB II
SUN/HP/AIX VM
zTCP
Blade/Linux snmp CP Monitor
(VSE!!!) LINUX
(VMWare) SNMP/Host Mibs

ESATCP: Network Monitor zMON Real-Time

 SNMP Data collection Displays
 Data added to PDB
Availability Checking PDB

zMAP Reports
Collects data from:
PDB
 LINUX (netsnmp)
 NT/SUN/HP (native snmp)
Printers/Routers….
27

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Data Collection Scalability

Using snmpd is VERY scalable!!!

Performance collection for 1000 servers: 30% of one IFL?
Report: ESALNXP LINUX HOST Process Statistics Report
---------------------------------------------------------
node/ <-Process Ident-> Nice <------CPU Percents---->
Name ID PPID GRP Valu Tot sys user syst usrt
--------- ----- ----- ----- ---- ---- ---- ---- ---- ----
snmpd 2290 1 2289 -10 0.02 0.02 0 0 0
snmpd 1776 1 1775 -10 0.02 0.02 0 0 0
snmpd 2069 1 2068 -10 0.02 0 0.02 0 0
snmpd 2193 1 2192 -10 0.25 0.12 0.12 0 0
snmpd 2166 1 2165 -10 0.04 0.02 0.02 0 0
snmpd 2230 1 2229 -10 0.04 0.02 0.02 0 0
snmpd 3370 1 3369 -10 0.03 0.02 0.02 0 0
snmpd 11202 1 11201 -10 0.04 0.02 0.02 0 0
snmpd 7855 1 7854 -10 0.02 0.02 0 0 0
snmpd 5848 1 5847 -10 0.08 0.06 0.02 0 0
snmpd 2290 1 2289 -10 0.04 0.02 0.02 0 0
snmpd 1776 1 1775 -10 0.04 0.02 0.02 0 0
snmpd 2069 1 2068 -10 0.02 0.02 0 0 0

28

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Drill down technology

Ability to see wide range of data, drill down

29

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Drill down technology

Drill down

30

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 1000 Server Disk Space Requirements

One minute performance data

• 4000 cylinders for 1000 servers per day
• 4 cylinders per server per day
Daily Archive
• 300 cylinders for 1,000 servers per day

Disk space “tailorable” with thresholds

31

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 VM “granularity” in zVPS

User classification (Virtual Machines)

• USERCLASS(3,x) = “HTTP*”
• USERCLASS(4,x) = “SAP*” ;sap servers
• USERCLASS(5,x) = “Z*” ;Velocity servers
Node classification
• USERCLASS(6,x) = “ESXA*” ; esx server
• USERCLASS(7,x) = “ESXB*”

32

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Operational Alert Support

TCPIP VM
SNMP/MIB II zTCP
CP Monitor

LINUX Alerts
SNMP/Host Mibs VM:Operator
zMON

PDB

zMAP

33

Copyright 2011 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis Education
Velocity Software’s performance workship
• (“free” in April)

VM Workshop – University of Kentucky

SHARE

User groups

34

Copyright 2010 Velocity Software, Inc. All Rights Reserved.

 Performance Analysis Summary
Analyzing 1000 virtual machines was easy 20 years ago

Analyzing today’s Linux servers requires added data

• Linux data
• Other virtual platforms

1000 (10,000) servers requires:

• Management.
• Scalable management tools
• No Crystal balls

35

Copyright 2010 Velocity Software, Inc. All Rights Reserved.