SlideShare a Scribd company logo
Scalable performance monitoring of networks, servers and
            applications using standard metrics
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Host Structures
Host Structures
• CPU: load_one, load_five, load_fifteen, proc_run,
  proc_total, cpu_num, cpu_speed, uptime, cpu_user,
   cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr,
   cpu_sintr, interupts, contexts
Host Structures
• CPU: load_one, load_five, load_fifteen, proc_run,
  proc_total, cpu_num, cpu_speed, uptime, cpu_user,
   cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr,
   cpu_sintr, interupts, contexts


• Memory: mem_total, mem_free, mem_shared,
  mem_buffers, mem_cached, swap_total, swap_free,
   page_in, page_out, swap_in, swap_out
Host Structures
• CPU: load_one, load_five, load_fifteen, proc_run,
  proc_total, cpu_num, cpu_speed, uptime, cpu_user,
   cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr,
   cpu_sintr, interupts, contexts


• Memory: mem_total, mem_free, mem_shared,
  mem_buffers, mem_cached, swap_total, swap_free,
   page_in, page_out, swap_in, swap_out


• Disk IO: disk_total, disk_free, part_max_used, reads,
  bytes_read, read_time, writes, bytes_written, write_time
Host Structures
• CPU: load_one, load_five, load_fifteen, proc_run,
  proc_total, cpu_num, cpu_speed, uptime, cpu_user,
   cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr,
   cpu_sintr, interupts, contexts


• Memory: mem_total, mem_free, mem_shared,
  mem_buffers, mem_cached, swap_total, swap_free,
   page_in, page_out, swap_in, swap_out


• Disk IO: disk_total, disk_free, part_max_used, reads,
  bytes_read, read_time, writes, bytes_written, write_time


• Network IO: bytes_in, packets_in, errs_in, drops_in,
   bytes_out, packet_out, errs_out, drops_out
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Standard measurements
Web Services
Web Services
• HTTP Counters: method_option_count,
  method_get_count, method_head_count,
  method_post_count, method_put_count,
  method_delete_count, method_trace_count,
  method_connect_count, method_other_count,
  status_1xx_count, status_2xx_count, status_3xx_count,
  status_4xx_count, status_5xx_count, status_other_count
Web Services
• HTTP Counters: method_option_count,
  method_get_count, method_head_count,
   method_post_count, method_put_count,
   method_delete_count, method_trace_count,
   method_connect_count, method_other_count,
   status_1xx_count, status_2xx_count, status_3xx_count,
   status_4xx_count, status_5xx_count, status_other_count


• HTTP Operations: method, uri, host, referer,
  useragent, authuser, mime-type, bytes, duration, status
Web Services
   • HTTP Counters: method_option_count,
     method_get_count, method_head_count,
      method_post_count, method_put_count,
      method_delete_count, method_trace_count,
      method_connect_count, method_other_count,
      status_1xx_count, status_2xx_count, status_3xx_count,
      status_4xx_count, status_5xx_count, status_other_count


   • HTTP Operations: method, uri, host, referer,
     useragent, authuser, mime-type, bytes, duration, status


sFlow supports random sampling of operations for
scalability - centralized monitoring of thousands of web
servers, load balancers etc.
Unified data model links network, servers and applications

                             Sampled Transactions

                             Transaction Counters
              APPLICATION
                               TCP/UDP Socket




                                    CPU

                                   Memory

                                     I/O
                 HOST
                                 Power, Temp.

                                Adapter MACs




                            Sampled Packet Headers

                                 I/F Counters
              NETWORK
                                 Power, Temp.
Unified data model links network, servers and applications

                             Sampled Transactions

                             Transaction Counters
              APPLICATION
                               TCP/UDP Socket




                                    CPU

                                   Memory                           Packet Header

                                     I/O                Source                      Destination
                 HOST
                                 Power, Temp.        TCP/UDP Socket             TCP/UDP Socket

                                Adapter MACs          MAC Address                   MAC Address




                            Sampled Packet Headers

                                 I/F Counters
              NETWORK
                                 Power, Temp.
Unified data model links network, servers and applications

                             Sampled Transactions

                             Transaction Counters
              APPLICATION
                               TCP/UDP Socket




                                    CPU

                                   Memory                           Packet Header

                                     I/O                Source                      Destination
                 HOST
                                 Power, Temp.        TCP/UDP Socket             TCP/UDP Socket

                                Adapter MACs          MAC Address                   MAC Address




                            Sampled Packet Headers

                                 I/F Counters
              NETWORK
                                 Power, Temp.
Unified data model links network, servers and applications

                             Sampled Transactions

                             Transaction Counters
              APPLICATION
                               TCP/UDP Socket




                                    CPU

                                   Memory                           Packet Header

                                     I/O                Source                      Destination
                 HOST
                                 Power, Temp.        TCP/UDP Socket             TCP/UDP Socket

                                Adapter MACs          MAC Address                   MAC Address




                            Sampled Packet Headers

                                 I/F Counters
              NETWORK
                                 Power, Temp.
Current Activities
Current Activities
• sFlow.org standardizing metrics for core services
  (HTTP, Memcache, HDFS, NFS etc)
Current Activities
• sFlow.org standardizing metrics for core services
  (HTTP, Memcache, HDFS, NFS etc)

• Embed sFlow in operating systems, hypervisors and
  applications (Apache, NGINX, HAproxy,
  Memcached, Membase, Hadoop ...)
Current Activities
• sFlow.org standardizing metrics for core services
  (HTTP, Memcache, HDFS, NFS etc)

• Embed sFlow in operating systems, hypervisors and
  applications (Apache, NGINX, HAproxy,
  Memcached, Membase, Hadoop ...)

• Native support for sFlow in performance
  monitoring tools (Ganglia, Nagios, Collectd, Munin,
  log file analyzers etc.)
Current Activities
• sFlow.org standardizing metrics for core services
  (HTTP, Memcache, HDFS, NFS etc)

• Embed sFlow in operating systems, hypervisors and
  applications (Apache, NGINX, HAproxy,
  Memcached, Membase, Hadoop ...)

• Native support for sFlow in performance
  monitoring tools (Ganglia, Nagios, Collectd, Munin,
  log file analyzers etc.)

• Integrated network, server and application visibility

More Related Content

What's hot (15)

PPT
Multi-Protocol Label Switching: Basics and Applications
Vishal Sharma, Ph.D.
 
PPTX
Mpls
arbhatawdekar
 
PDF
Deploy MPLS Traffic Engineering
APNIC
 
PPT
Mpls Services
Kristof De Brouwer
 
PDF
MPLS Deployment Chapter 1 - Basic
Ericsson
 
PPTX
MPLS L3 VPN Tutorial, by Nurul Islam Roman [APNIC 38]
APNIC
 
PPT
Mpls vpn toi
ARACELY ZAPATA
 
PPT
Juniper mpls best practice part 2
Febrian ‎
 
PDF
MPLS Deployment Chapter 2 - Services
Ericsson
 
PDF
MPLS Concepts and Fundamentals
Shawn Zandi
 
PDF
Service Density By Xelerated At Linley Seminar
Xelerated
 
PDF
Tma ph d_school_2011
Muhammad Wasif
 
PDF
Network Processing on an SPE Core in Cell Broadband EngineTM
Slide_N
 
PDF
Juniper MPLS Tutorial by Soricelli
Febrian ‎
 
Multi-Protocol Label Switching: Basics and Applications
Vishal Sharma, Ph.D.
 
Deploy MPLS Traffic Engineering
APNIC
 
Mpls Services
Kristof De Brouwer
 
MPLS Deployment Chapter 1 - Basic
Ericsson
 
MPLS L3 VPN Tutorial, by Nurul Islam Roman [APNIC 38]
APNIC
 
Mpls vpn toi
ARACELY ZAPATA
 
Juniper mpls best practice part 2
Febrian ‎
 
MPLS Deployment Chapter 2 - Services
Ericsson
 
MPLS Concepts and Fundamentals
Shawn Zandi
 
Service Density By Xelerated At Linley Seminar
Xelerated
 
Tma ph d_school_2011
Muhammad Wasif
 
Network Processing on an SPE Core in Cell Broadband EngineTM
Slide_N
 
Juniper MPLS Tutorial by Soricelli
Febrian ‎
 

Viewers also liked (15)

PDF
SD - A peer to peer issue tracking system
Jesse Vincent
 
PDF
The Carrier DevOps Trend (Presented to Okinawa Open Days Conference)
Alex Henthorn-Iwane
 
PPTX
Swarm - A Docker Clustering System
snrism
 
PDF
Next-Gen DDoS Detection
Alex Henthorn-Iwane
 
PDF
Swarm sec
snrism
 
PDF
Cloud Aware Network Management
Alex Henthorn-Iwane
 
PPTX
SDN Controller - Programming Challenges
snrism
 
PDF
Cloud-Scale BGP and NetFlow Analysis
Alex Henthorn-Iwane
 
PDF
5G-USA-Telemetry
snrism
 
PDF
垂直互联网站点的技术改造
Dahui Feng
 
PDF
Devtest Orchestration for SDN & NFV
Alex Henthorn-Iwane
 
PDF
Docker-OVS
snrism
 
PDF
Edge architecture ieee international conference on cloud engineering
Mikey Cohen - Hiring Amazing Engineers
 
PPTX
Docker networking Tutorial 101
LorisPack Project
 
PPTX
Zuul @ Netflix SpringOne Platform
Mikey Cohen - Hiring Amazing Engineers
 
SD - A peer to peer issue tracking system
Jesse Vincent
 
The Carrier DevOps Trend (Presented to Okinawa Open Days Conference)
Alex Henthorn-Iwane
 
Swarm - A Docker Clustering System
snrism
 
Next-Gen DDoS Detection
Alex Henthorn-Iwane
 
Swarm sec
snrism
 
Cloud Aware Network Management
Alex Henthorn-Iwane
 
SDN Controller - Programming Challenges
snrism
 
Cloud-Scale BGP and NetFlow Analysis
Alex Henthorn-Iwane
 
5G-USA-Telemetry
snrism
 
垂直互联网站点的技术改造
Dahui Feng
 
Devtest Orchestration for SDN & NFV
Alex Henthorn-Iwane
 
Docker-OVS
snrism
 
Edge architecture ieee international conference on cloud engineering
Mikey Cohen - Hiring Amazing Engineers
 
Docker networking Tutorial 101
LorisPack Project
 
Zuul @ Netflix SpringOne Platform
Mikey Cohen - Hiring Amazing Engineers
 
Ad

Similar to Standard measurements (20)

PPT
slides
webhostingguy
 
PDF
High perf-networking
mtimjones
 
KEY
The sFlow Standard: Scalable, Unified Monitoring of Networks, Systems and App...
netvis
 
PPT
Application Layer and Socket Programming
elliando dias
 
PPTX
Driver Configuration Webinar
AVEVA
 
PPTX
FlowER Erlang Openflow Controller
Holger Winkelmann
 
PDF
10 Reasons to use the Renesas R-IN multi-protocol industrial ethernet solutio...
Renesas Electronics Corporation
 
PPTX
InduSoft Driver Configuration Webinar
AVEVA
 
PPTX
Iot platform supporting million requests per second
Abinasha Karana
 
PDF
CoAP, Copper, and Embedded Web Resources
Matthias Kovatsch
 
PPTX
Tcp and introduction to protocol
Sripati Mahapatra
 
PPTX
Tcpandintroductiontoprotocol 150618054958-lva1-app6892
Saumendra Pradhan
 
PPTX
Wireshark, Tcpdump and Network Performance tools
Sachidananda Sahu
 
PPT
App layer
khushali_modi
 
PDF
Basic ccna interview questions and answers ~ sysnet notes
Vamsi Krishna Kalavala
 
PDF
Ajp notes-chapter-04
Ankit Dubey
 
PPTX
Sanitizing PCAPs
Jasper Bongertz
 
PPSX
Certified Hospitality Technology Professional
Huy Pham
 
PPT
COMPUTER NETWORKS
Manjula Pradeep Gunathilake
 
PPT
07 Input Output
Jeanie Delos Arcos
 
High perf-networking
mtimjones
 
The sFlow Standard: Scalable, Unified Monitoring of Networks, Systems and App...
netvis
 
Application Layer and Socket Programming
elliando dias
 
Driver Configuration Webinar
AVEVA
 
FlowER Erlang Openflow Controller
Holger Winkelmann
 
10 Reasons to use the Renesas R-IN multi-protocol industrial ethernet solutio...
Renesas Electronics Corporation
 
InduSoft Driver Configuration Webinar
AVEVA
 
Iot platform supporting million requests per second
Abinasha Karana
 
CoAP, Copper, and Embedded Web Resources
Matthias Kovatsch
 
Tcp and introduction to protocol
Sripati Mahapatra
 
Tcpandintroductiontoprotocol 150618054958-lva1-app6892
Saumendra Pradhan
 
Wireshark, Tcpdump and Network Performance tools
Sachidananda Sahu
 
App layer
khushali_modi
 
Basic ccna interview questions and answers ~ sysnet notes
Vamsi Krishna Kalavala
 
Ajp notes-chapter-04
Ankit Dubey
 
Sanitizing PCAPs
Jasper Bongertz
 
Certified Hospitality Technology Professional
Huy Pham
 
COMPUTER NETWORKS
Manjula Pradeep Gunathilake
 
07 Input Output
Jeanie Delos Arcos
 
Ad

Recently uploaded (20)

PPTX
Introduction to Flutter by Ayush Desai.pptx
ayushdesai204
 
PDF
NewMind AI Weekly Chronicles – July’25, Week III
NewMind AI
 
PPTX
The Future of AI & Machine Learning.pptx
pritsen4700
 
PDF
RAT Builders - How to Catch Them All [DeepSec 2024]
malmoeb
 
PDF
CIFDAQ's Market Wrap : Bears Back in Control?
CIFDAQ
 
PPTX
What-is-the-World-Wide-Web -- Introduction
tonifi9488
 
PDF
TrustArc Webinar - Navigating Data Privacy in LATAM: Laws, Trends, and Compli...
TrustArc
 
PDF
Research-Fundamentals-and-Topic-Development.pdf
ayesha butalia
 
PDF
Brief History of Internet - Early Days of Internet
sutharharshit158
 
PDF
The Future of Mobile Is Context-Aware—Are You Ready?
iProgrammer Solutions Private Limited
 
PDF
Presentation about Hardware and Software in Computer
snehamodhawadiya
 
PDF
Market Insight : ETH Dominance Returns
CIFDAQ
 
PPTX
AI and Robotics for Human Well-being.pptx
JAYMIN SUTHAR
 
PPTX
Farrell_Programming Logic and Design slides_10e_ch02_PowerPoint.pptx
bashnahara11
 
PDF
Responsible AI and AI Ethics - By Sylvester Ebhonu
Sylvester Ebhonu
 
PDF
Generative AI vs Predictive AI-The Ultimate Comparison Guide
Lily Clark
 
PPTX
IT Runs Better with ThousandEyes AI-driven Assurance
ThousandEyes
 
PDF
Build with AI and GDG Cloud Bydgoszcz- ADK .pdf
jaroslawgajewski1
 
PPTX
Simple and concise overview about Quantum computing..pptx
mughal641
 
PDF
Data_Analytics_vs_Data_Science_vs_BI_by_CA_Suvidha_Chaplot.pdf
CA Suvidha Chaplot
 
Introduction to Flutter by Ayush Desai.pptx
ayushdesai204
 
NewMind AI Weekly Chronicles – July’25, Week III
NewMind AI
 
The Future of AI & Machine Learning.pptx
pritsen4700
 
RAT Builders - How to Catch Them All [DeepSec 2024]
malmoeb
 
CIFDAQ's Market Wrap : Bears Back in Control?
CIFDAQ
 
What-is-the-World-Wide-Web -- Introduction
tonifi9488
 
TrustArc Webinar - Navigating Data Privacy in LATAM: Laws, Trends, and Compli...
TrustArc
 
Research-Fundamentals-and-Topic-Development.pdf
ayesha butalia
 
Brief History of Internet - Early Days of Internet
sutharharshit158
 
The Future of Mobile Is Context-Aware—Are You Ready?
iProgrammer Solutions Private Limited
 
Presentation about Hardware and Software in Computer
snehamodhawadiya
 
Market Insight : ETH Dominance Returns
CIFDAQ
 
AI and Robotics for Human Well-being.pptx
JAYMIN SUTHAR
 
Farrell_Programming Logic and Design slides_10e_ch02_PowerPoint.pptx
bashnahara11
 
Responsible AI and AI Ethics - By Sylvester Ebhonu
Sylvester Ebhonu
 
Generative AI vs Predictive AI-The Ultimate Comparison Guide
Lily Clark
 
IT Runs Better with ThousandEyes AI-driven Assurance
ThousandEyes
 
Build with AI and GDG Cloud Bydgoszcz- ADK .pdf
jaroslawgajewski1
 
Simple and concise overview about Quantum computing..pptx
mughal641
 
Data_Analytics_vs_Data_Science_vs_BI_by_CA_Suvidha_Chaplot.pdf
CA Suvidha Chaplot
 

Standard measurements

  • 1. Scalable performance monitoring of networks, servers and applications using standard metrics
  • 25. Host Structures • CPU: load_one, load_five, load_fifteen, proc_run, proc_total, cpu_num, cpu_speed, uptime, cpu_user, cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr, cpu_sintr, interupts, contexts
  • 26. Host Structures • CPU: load_one, load_five, load_fifteen, proc_run, proc_total, cpu_num, cpu_speed, uptime, cpu_user, cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr, cpu_sintr, interupts, contexts • Memory: mem_total, mem_free, mem_shared, mem_buffers, mem_cached, swap_total, swap_free, page_in, page_out, swap_in, swap_out
  • 27. Host Structures • CPU: load_one, load_five, load_fifteen, proc_run, proc_total, cpu_num, cpu_speed, uptime, cpu_user, cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr, cpu_sintr, interupts, contexts • Memory: mem_total, mem_free, mem_shared, mem_buffers, mem_cached, swap_total, swap_free, page_in, page_out, swap_in, swap_out • Disk IO: disk_total, disk_free, part_max_used, reads, bytes_read, read_time, writes, bytes_written, write_time
  • 28. Host Structures • CPU: load_one, load_five, load_fifteen, proc_run, proc_total, cpu_num, cpu_speed, uptime, cpu_user, cpu_nice, cpu_system, cpu_idle, cpu_wio, cpu_intr, cpu_sintr, interupts, contexts • Memory: mem_total, mem_free, mem_shared, mem_buffers, mem_cached, swap_total, swap_free, page_in, page_out, swap_in, swap_out • Disk IO: disk_total, disk_free, part_max_used, reads, bytes_read, read_time, writes, bytes_written, write_time • Network IO: bytes_in, packets_in, errs_in, drops_in, bytes_out, packet_out, errs_out, drops_out
  • 43. Web Services • HTTP Counters: method_option_count, method_get_count, method_head_count, method_post_count, method_put_count, method_delete_count, method_trace_count, method_connect_count, method_other_count, status_1xx_count, status_2xx_count, status_3xx_count, status_4xx_count, status_5xx_count, status_other_count
  • 44. Web Services • HTTP Counters: method_option_count, method_get_count, method_head_count, method_post_count, method_put_count, method_delete_count, method_trace_count, method_connect_count, method_other_count, status_1xx_count, status_2xx_count, status_3xx_count, status_4xx_count, status_5xx_count, status_other_count • HTTP Operations: method, uri, host, referer, useragent, authuser, mime-type, bytes, duration, status
  • 45. Web Services • HTTP Counters: method_option_count, method_get_count, method_head_count, method_post_count, method_put_count, method_delete_count, method_trace_count, method_connect_count, method_other_count, status_1xx_count, status_2xx_count, status_3xx_count, status_4xx_count, status_5xx_count, status_other_count • HTTP Operations: method, uri, host, referer, useragent, authuser, mime-type, bytes, duration, status sFlow supports random sampling of operations for scalability - centralized monitoring of thousands of web servers, load balancers etc.
  • 46. Unified data model links network, servers and applications Sampled Transactions Transaction Counters APPLICATION TCP/UDP Socket CPU Memory I/O HOST Power, Temp. Adapter MACs Sampled Packet Headers I/F Counters NETWORK Power, Temp.
  • 47. Unified data model links network, servers and applications Sampled Transactions Transaction Counters APPLICATION TCP/UDP Socket CPU Memory Packet Header I/O Source Destination HOST Power, Temp. TCP/UDP Socket TCP/UDP Socket Adapter MACs MAC Address MAC Address Sampled Packet Headers I/F Counters NETWORK Power, Temp.
  • 48. Unified data model links network, servers and applications Sampled Transactions Transaction Counters APPLICATION TCP/UDP Socket CPU Memory Packet Header I/O Source Destination HOST Power, Temp. TCP/UDP Socket TCP/UDP Socket Adapter MACs MAC Address MAC Address Sampled Packet Headers I/F Counters NETWORK Power, Temp.
  • 49. Unified data model links network, servers and applications Sampled Transactions Transaction Counters APPLICATION TCP/UDP Socket CPU Memory Packet Header I/O Source Destination HOST Power, Temp. TCP/UDP Socket TCP/UDP Socket Adapter MACs MAC Address MAC Address Sampled Packet Headers I/F Counters NETWORK Power, Temp.
  • 51. Current Activities • sFlow.org standardizing metrics for core services (HTTP, Memcache, HDFS, NFS etc)
  • 52. Current Activities • sFlow.org standardizing metrics for core services (HTTP, Memcache, HDFS, NFS etc) • Embed sFlow in operating systems, hypervisors and applications (Apache, NGINX, HAproxy, Memcached, Membase, Hadoop ...)
  • 53. Current Activities • sFlow.org standardizing metrics for core services (HTTP, Memcache, HDFS, NFS etc) • Embed sFlow in operating systems, hypervisors and applications (Apache, NGINX, HAproxy, Memcached, Membase, Hadoop ...) • Native support for sFlow in performance monitoring tools (Ganglia, Nagios, Collectd, Munin, log file analyzers etc.)
  • 54. Current Activities • sFlow.org standardizing metrics for core services (HTTP, Memcache, HDFS, NFS etc) • Embed sFlow in operating systems, hypervisors and applications (Apache, NGINX, HAproxy, Memcached, Membase, Hadoop ...) • Native support for sFlow in performance monitoring tools (Ganglia, Nagios, Collectd, Munin, log file analyzers etc.) • Integrated network, server and application visibility

Editor's Notes

  • #2: Hi. My name is Peter Phaal and I am one of the authors of the sFlow standard. \n\nIn this talk we will be looking at some of the scalability challenges in current performance monitoring approaches and explore how standard performance metrics can enhance the scalability of network, server and application monitoring systems.\n\nFirst, let me quickly briefly introduce sFlow.\n
  • #3: sFlow is a multi-vendor industry standard for monitoring performance in high-speed switched networks. The standard identifies a common set of metrics and provides an efficient way to centrally monitor network performance.\n
  • #4: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #5: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #6: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #7: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #8: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #9: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #10: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #11: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #12: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #13: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #14: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #15: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #16: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #17: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #18: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #19: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #20: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #21: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #22: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #23: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #24: sFlow is widely supported by network vendors, allowing network managers to choose from a wide variety of open source and commercial performance management applications.\n\nVirtualization and scale-out applications are transforming network architectures, closely linking network and system performance. In order to address this challenge, sFlow has been extended to include server performance monitoring. \n\nTo understand how sFlow can help simplify performance management and improve scalability, it is worth looking at current approaches to server performance monitoring.\n\n\n
  • #25: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #26: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #27: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #28: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #29: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #30: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #31: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #32: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #33: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #34: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #35: Let’s start with something simple, monitoring KVM servers using Cacti. Setting up the monitoring system involves installing agents on the servers and configuring Cacti to collect data. Next you decide that you would also like to monitor additional platforms, including: Linux, Xen, Citrix XenServer and Windows. Each new platform requires an agent and additional configuration before it can be monitored.\n\nAfter a while you decide to deploy Nagios to monitor status and generate events. This involves additional agents for each platform and additional configuration to monitor each server. Next Ganglia is installed to monitor the performance of server clusters, requiring additional agents and configuration. Each new tool provides distinct capabilities, but at the cost of operational complexity and overhead associated with deploying additional agents.\n
  • #36: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #37: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #38: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #39: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #40: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #41: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #42: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #43: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #44: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #45: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #46: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #47: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #48: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #49: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #50: Looking at each of the performance monitoring tools in a little more detail. This is a chart showing CPU utilization in:\nGanglia, \nNagios, \nCacti, \nCollectd \nand Munin. No prizes for spotting the similarities between the charts.\n\nIt is clear that there is a common set of metrics that is generally accepted for system performance monitoring and that these metrics deliver the core functionality in performance monitoring tools. It is these core statistics that are now part of the sFlow standard.\n
  • #51: The standard sFlow metrics cover \nCPU, \nMemory, \nDisk IO and \nNetwork IO. \n\nFor anyone familiar with Ganglia, these metrics should be instantly recognizable. The choice of metrics was heavily influenced by the Ganglia project which has a developed a mature set set of core metrics that can be gathered from a wide range of operation systems.\n
  • #52: The standard sFlow metrics cover \nCPU, \nMemory, \nDisk IO and \nNetwork IO. \n\nFor anyone familiar with Ganglia, these metrics should be instantly recognizable. The choice of metrics was heavily influenced by the Ganglia project which has a developed a mature set set of core metrics that can be gathered from a wide range of operation systems.\n
  • #53: The standard sFlow metrics cover \nCPU, \nMemory, \nDisk IO and \nNetwork IO. \n\nFor anyone familiar with Ganglia, these metrics should be instantly recognizable. The choice of metrics was heavily influenced by the Ganglia project which has a developed a mature set set of core metrics that can be gathered from a wide range of operation systems.\n
  • #54: The standard sFlow metrics cover \nCPU, \nMemory, \nDisk IO and \nNetwork IO. \n\nFor anyone familiar with Ganglia, these metrics should be instantly recognizable. The choice of metrics was heavily influenced by the Ganglia project which has a developed a mature set set of core metrics that can be gathered from a wide range of operation systems.\n
  • #55: The Host sFlow project is hosted on Sourceforge and provides an open source agent implementing the sFlow standard. \n\nThe agent currently runs on \nLinux, \nFreeBSD \nand Windows systems, \n\nalong with KVM \nand Xen based hypervisors.\n
  • #56: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #57: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #58: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #59: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #60: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #61: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #62: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #63: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #64: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #65: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #66: Lets see how standard metrics transform performance monitoring. In this case, sFlow agents can be deployed as a standard component of each platform.\n\nDeploying a performance monitoring application simply involves turning on the already deployed agents. Additional management tools can share the same standard measurement stream so there is no additional overhead involved in adding applications. Standard metrics ensure consistency between measurements displayed on the different tools making it easier to transition between tools.\n\nStandard agents allow performance monitoring tools to focus on data analysis and presentation, rather than the complexity of maintaining agents on different platforms. Pooling resources on the agent side eliminates wasteful duplication and greatly simplifies operational deployment.\n\nIn addition, standard agents lower the barrier to entry, making it easier to create and deploy innovative performance management tools.\n
  • #67: Another area that’s ripe for standardization is web services. Current efforts in sFlow.org are focussing on the key metrics that can describe the performance of web servers, load balancers etc.\n\nA standard set of counters is only part of the story. The sFlow protocol also provides a way to monitor HTTP operations. Defining a standard set of transaction attributes allows sFlow to efficiently report on web activity across larger server farms. Here again, there is already broad consensus on the set of attributes. The combined log file format is widely supported. Exporting the same data using sFlow provides real-time, centralized monitoring and can easily be converted back into a log file for existing tools as well as allowing a new class of real-time log analyzers to be created for performance monitoring.\n\nThe sFlow standard also support statistical sampling. By randomly selecting 1 in N records, you can centrally monitor the largest of web services.\n
  • #68: Another area that’s ripe for standardization is web services. Current efforts in sFlow.org are focussing on the key metrics that can describe the performance of web servers, load balancers etc.\n\nA standard set of counters is only part of the story. The sFlow protocol also provides a way to monitor HTTP operations. Defining a standard set of transaction attributes allows sFlow to efficiently report on web activity across larger server farms. Here again, there is already broad consensus on the set of attributes. The combined log file format is widely supported. Exporting the same data using sFlow provides real-time, centralized monitoring and can easily be converted back into a log file for existing tools as well as allowing a new class of real-time log analyzers to be created for performance monitoring.\n\nThe sFlow standard also support statistical sampling. By randomly selecting 1 in N records, you can centrally monitor the largest of web services.\n
  • #69: Another area that’s ripe for standardization is web services. Current efforts in sFlow.org are focussing on the key metrics that can describe the performance of web servers, load balancers etc.\n\nA standard set of counters is only part of the story. The sFlow protocol also provides a way to monitor HTTP operations. Defining a standard set of transaction attributes allows sFlow to efficiently report on web activity across larger server farms. Here again, there is already broad consensus on the set of attributes. The combined log file format is widely supported. Exporting the same data using sFlow provides real-time, centralized monitoring and can easily be converted back into a log file for existing tools as well as allowing a new class of real-time log analyzers to be created for performance monitoring.\n\nThe sFlow standard also support statistical sampling. By randomly selecting 1 in N records, you can centrally monitor the largest of web services.\n
  • #70: Looking at the big picture, sFlow doesn’t just define performance metrics and a scalable method for collecting them.\n\nAn important part of the standard is a data model that links network, host and application metrics together. For example, when exporting host metrics the host also exports its MAC addresses, allowing network and server performance to be linked. When an application transaction is exported, the layer 4 socket associated with the transaction is also included, providing a way to link network and application performance.\n
  • #71: Looking at the big picture, sFlow doesn’t just define performance metrics and a scalable method for collecting them.\n\nAn important part of the standard is a data model that links network, host and application metrics together. For example, when exporting host metrics the host also exports its MAC addresses, allowing network and server performance to be linked. When an application transaction is exported, the layer 4 socket associated with the transaction is also included, providing a way to link network and application performance.\n
  • #72: Looking at the big picture, sFlow doesn’t just define performance metrics and a scalable method for collecting them.\n\nAn important part of the standard is a data model that links network, host and application metrics together. For example, when exporting host metrics the host also exports its MAC addresses, allowing network and server performance to be linked. When an application transaction is exported, the layer 4 socket associated with the transaction is also included, providing a way to link network and application performance.\n
  • #73: I would like to finish by describing current areas of activity and invite participation from the broader comminity.\n\nI’ve already touched on some of the areas where standard metrics are being defined. Other areas include memcache, Hadoop and networked storage.\n\nAs the metrics become settled, they need to be implemented within the different operating systems, hypervisors and applications so that they can be efficiently deployed in production environments.\n\nNative support for sFlow in existing performance management applications is needed to deliver value from the instrumentation.\n\nFinally, a single measurement system linking network, server and application performance allows a new class of application to be developed, simplifying performance management by providing an integrated view of performance.\n\n\n
  • #74: I would like to finish by describing current areas of activity and invite participation from the broader comminity.\n\nI’ve already touched on some of the areas where standard metrics are being defined. Other areas include memcache, Hadoop and networked storage.\n\nAs the metrics become settled, they need to be implemented within the different operating systems, hypervisors and applications so that they can be efficiently deployed in production environments.\n\nNative support for sFlow in existing performance management applications is needed to deliver value from the instrumentation.\n\nFinally, a single measurement system linking network, server and application performance allows a new class of application to be developed, simplifying performance management by providing an integrated view of performance.\n\n\n
  • #75: I would like to finish by describing current areas of activity and invite participation from the broader comminity.\n\nI’ve already touched on some of the areas where standard metrics are being defined. Other areas include memcache, Hadoop and networked storage.\n\nAs the metrics become settled, they need to be implemented within the different operating systems, hypervisors and applications so that they can be efficiently deployed in production environments.\n\nNative support for sFlow in existing performance management applications is needed to deliver value from the instrumentation.\n\nFinally, a single measurement system linking network, server and application performance allows a new class of application to be developed, simplifying performance management by providing an integrated view of performance.\n\n\n
  • #76: I would like to finish by describing current areas of activity and invite participation from the broader comminity.\n\nI’ve already touched on some of the areas where standard metrics are being defined. Other areas include memcache, Hadoop and networked storage.\n\nAs the metrics become settled, they need to be implemented within the different operating systems, hypervisors and applications so that they can be efficiently deployed in production environments.\n\nNative support for sFlow in existing performance management applications is needed to deliver value from the instrumentation.\n\nFinally, a single measurement system linking network, server and application performance allows a new class of application to be developed, simplifying performance management by providing an integrated view of performance.\n\n\n