80 likes | 263 Views
E N D
1. Making the Grade: Ensuring Application Performance in an Education Network
Presented By:
Sean Applegate
Mid-Atlantic Systems Engineer
2. Slide 2 The Bottleneck Where is this most apparent? Where there’s the least amount of bandwidth--
The WAN Access Link
On either side of the WAN Access link, there’s lot of bandwidth
High-speed Internet backbone
High-speed corporate LANs
In between, there’s a low-speed link -- the WAN access link --
HUGE point of frustration
But whenever there’s speed disparity in the network, there are performance problems..
The problem is evident when traffic on the high-speed LAN (100 Mbps) hits the low-speed WAN (1.5 Mbps).Where is this most apparent? Where there’s the least amount of bandwidth--
The WAN Access Link
On either side of the WAN Access link, there’s lot of bandwidth
High-speed Internet backbone
High-speed corporate LANs
In between, there’s a low-speed link -- the WAN access link --
HUGE point of frustration
But whenever there’s speed disparity in the network, there are performance problems..
The problem is evident when traffic on the high-speed LAN (100 Mbps) hits the low-speed WAN (1.5 Mbps).
3. Slide 3 Introductory Questions How many people here own a PacketShaper?
How many people here are evaluating a PacketShaper?
How many people have NEVER heard of Packeteer or the PacketShaper?
How many people don’t know what your top 10 applications are and the percent of bandwidth they are using?
How many people are considering increasing WAN bandwidth speeds?
How many people are using other bandwidth mgnt or policing technologies to control traffic?
4. Slide 4 Applications Drive Today’s Educational Institutions Mission-critical applications are critical to education
All traffic is not created equal Today mission-critical applications such as SAP and Oracle are the crux of competitiveness. Educational institutions are spending a huge portion of their IT budgets on these expensive, mission-critical applications only to find them struggling to function efficiently over the network. Today mission-critical applications such as SAP and Oracle are the crux of competitiveness. Educational institutions are spending a huge portion of their IT budgets on these expensive, mission-critical applications only to find them struggling to function efficiently over the network.
5. Slide 5 What am I spending my $ on?
6. Slide 6 Agenda Who is Packeteer?
What is PacketShaper?
Who is using PacketShaper?
Implementing Packeteer's Four Step Process
- Classify
- Analyze
- Control
- Report
Summary
Questions
7. Slide 7 Packeteer Fact Sheet Founded in 1996, Pioneer of Proactive Bandwidth Management
Headquarters in Cupertino, CA
US Offices: New Jersey, Chicago, Atlanta, Dallas, Seattle, Washington D.C., San Diego
Offices Abroad: Netherlands, Hong Kong, Japan, and Australia
Regional Resellers
Employees: 197
Customer proven
Shipping since February 1997
Thousands of PacketShapers shipped worldwide
5th generation of software Packeteer was founded in 1996 to develop intelligent bandwidth management products. We launched the company and our first product in Fall 1996 at Internet World New York. Since that time we’ve grown the company to upwards of 120 people - extending the product line, development team, and opening up sales offices worldwide including our European headquarters in the Netherlands and Asian headquarters in Hong Kong, plus offices in Japan and Australia.
Packeteer was founded in 1996 to develop intelligent bandwidth management products. We launched the company and our first product in Fall 1996 at Internet World New York. Since that time we’ve grown the company to upwards of 120 people - extending the product line, development team, and opening up sales offices worldwide including our European headquarters in the Netherlands and Asian headquarters in Hong Kong, plus offices in Japan and Australia.
8. Slide 8 Packeteer Product Family Packeteer’s® PacketShaper™, AppVantage™, and AppCelera™ product lines provide today’s best solutions for application QoS and application-based services.
PacketShaper, Packeteer’s enterprise-and ISP-focused product line, is a bandwidth-management solution that brings predictable, efficient performance to Internet-enabled applications. The award-winning PacketShaper, deployed globally by thousands of customers since shipments began in early 1997, presents insightful analyses of application and network behavior and enforces policy-based bandwidth allocation, ensuring end-to-end application QoS.
AppVantage, Packeteer’s ASP-focused product line, is an application subscriber management solution that enables Application Service Providers to provision, monitor, measure, control, and validate an extensive portfolio of application services. It integrates service-level management, end-to-end quality of service, subscriber-ASP mediation points, and reporting into a unified, policy-driven system.
AppCelera, Packeteer’s newest product line, is for both enterprises and service providers. Its suite of unique technologies focus on acceleration of Internet applications and content. AppCelera ISX-50 Internet Security Accelerator, speeds secure e-business applications by handling performance-impacting SSL transactions on a dedicated platform and freeing overloaded web servers. AppCelera ICX-55 Internet Content Accelerator uses caching, compression, and transformation techniques to optimize content transfer for users’ browsers and connection speeds.
PolicyCenter, Packeteer’s directory-enabled NT-based solution for cost-effective policy management of PacketWise-enabled devices, greatly simplifies deployment of multiple PacketShapers or AppVantage appliances by centralizing policy and software-upgrade deployment and providing a summary view of all managed devices.
Packeteer’s® PacketShaper™, AppVantage™, and AppCelera™ product lines provide today’s best solutions for application QoS and application-based services.
PacketShaper, Packeteer’s enterprise-and ISP-focused product line, is a bandwidth-management solution that brings predictable, efficient performance to Internet-enabled applications. The award-winning PacketShaper, deployed globally by thousands of customers since shipments began in early 1997, presents insightful analyses of application and network behavior and enforces policy-based bandwidth allocation, ensuring end-to-end application QoS.
AppVantage, Packeteer’s ASP-focused product line, is an application subscriber management solution that enables Application Service Providers to provision, monitor, measure, control, and validate an extensive portfolio of application services. It integrates service-level management, end-to-end quality of service, subscriber-ASP mediation points, and reporting into a unified, policy-driven system.
AppCelera, Packeteer’s newest product line, is for both enterprises and service providers. Its suite of unique technologies focus on acceleration of Internet applications and content. AppCelera ISX-50 Internet Security Accelerator, speeds secure e-business applications by handling performance-impacting SSL transactions on a dedicated platform and freeing overloaded web servers. AppCelera ICX-55 Internet Content Accelerator uses caching, compression, and transformation techniques to optimize content transfer for users’ browsers and connection speeds.
PolicyCenter, Packeteer’s directory-enabled NT-based solution for cost-effective policy management of PacketWise-enabled devices, greatly simplifies deployment of multiple PacketShapers or AppVantage appliances by centralizing policy and software-upgrade deployment and providing a summary view of all managed devices.
9. Slide 9 What is PacketShaper? Industry Leading QoS Solution
CMP Network Telecom: Network Infrastructure Product of the Year - 2001
Firmware, Real-time OS (PSOS)
Classifies 350+ Apps at OSI Layers 2-7
Uses Patented TCP Rate Control to proactively control application traffic and prevent queuing and reduce latency.
Over 55 measurement variables for detailed analysis
Managed through an onboard web interface and CLI, no external hardware/software required
Becomes a piece of wire if it fails
PacketShaper lets you manage the allocation of bandwidth to different applications for optimal performance. The goal is to assign enough bandwidth to the mission-critical, revenue-generating applications to guarantee Quality of Service (QoS) to the application users. QoS is achieved when the network transports data without losing packets and provides predictable, appropriate end-to-end delay.
Let’s look at how PacketShaper helps you achieve this goal.
PacketShaper lets you manage the allocation of bandwidth to different applications for optimal performance. The goal is to assign enough bandwidth to the mission-critical, revenue-generating applications to guarantee Quality of Service (QoS) to the application users. QoS is achieved when the network transports data without losing packets and provides predictable, appropriate end-to-end delay.
Let’s look at how PacketShaper helps you achieve this goal.
10. Slide 10 PacketShaper Product Line PacketShaper is a software/hardware solution that sits on the LAN side of the router. Depending on the WAN link size, you would choose the PacketShaper 1500, 2500, 4500, or 6500 platform.
Installation of PacketShaper is simple -- Ethernet in, Ethernet out; deployed transparently between the LAN and the WAN access router, with no negative impact on any other products on the network. And PacketShaper’s design prevents it from being a point of failure on the network.
1500: Four entry points for WAN capacity, with six WAN capacity upgrade options; low profile unit – 1U height.
2500: Three entry points for WAN capacity, with three WAN capacity upgrade options; modular unit -- two PCI slots.
4500: Three entry points (one optimized for ISPs), with one WAN capacity upgrade option; modular unit -- two PCI slots; dual power sourcing.
6500: Three entry points (one optimized for ISPs), with one WAN capacity upgrade option; modular unit – two PCI slots; dual power sourcing.
PacketShaper is a software/hardware solution that sits on the LAN side of the router. Depending on the WAN link size, you would choose the PacketShaper 1500, 2500, 4500, or 6500 platform.
Installation of PacketShaper is simple -- Ethernet in, Ethernet out; deployed transparently between the LAN and the WAN access router, with no negative impact on any other products on the network. And PacketShaper’s design prevents it from being a point of failure on the network.
1500: Four entry points for WAN capacity, with six WAN capacity upgrade options; low profile unit – 1U height.
2500: Three entry points for WAN capacity, with three WAN capacity upgrade options; modular unit -- two PCI slots.
4500: Three entry points (one optimized for ISPs), with one WAN capacity upgrade option; modular unit -- two PCI slots; dual power sourcing.
6500: Three entry points (one optimized for ISPs), with one WAN capacity upgrade option; modular unit – two PCI slots; dual power sourcing.
11. Slide 11 Typical k-12 School District Topology
12. Slide 12 380+ Higher Education Customers in US Stanford Univ
California Tech
Yale Univ
Vanderbilt Univ
Univ of Miami
Texas A&M
Clemson Univ
Univ of Notre Dame
All Universities of California
Case Western Reserve Univ
Ohio Northern Univ
Dartmouth College
Howard Univ
Univ of Dayton
Miami Univ
Case Western Reserver Univ
Cleveland State Univ
Xavier Univ
Youngstown State
Denison Univ
Bowling Green Univ
Capital Univ
…Ohio has more higher ed users than any other state in the US.
…A Sites are catching up fast!
Thousands of customers worldwide are presently using PacketShaper.
Chemical and Petroleum:
Borden Chemical uses PacketShaper to protect SAP/R3 performance.
Cytec protects System Software Associates (SSA) BPCS performance.
Financial Services:
Grant Thornton extends the useful life of its current WAN bandwidth.
Northwestern Mutual Life enforces TN3270 response time service-level agreements.
Information Management:
RR Donnelly uses PacketShaper to ensure satisfactory Citrix performance across its WAN.
Standard & Poor’s protects mission-critical Oracle performance.
Service Providers:
Clear Communications uses PacketShaper to offer tiered services to its customers.
Technology:
Autodesk uses PacketShaper to protect SAP and contain Microsoft Exchange traffic.
Lucent Technology protects mission-critical telnet traffic traversing expensive international WAN connections.
Thousands of customers worldwide are presently using PacketShaper.
Chemical and Petroleum:
Borden Chemical uses PacketShaper to protect SAP/R3 performance.
Cytec protects System Software Associates (SSA) BPCS performance.
Financial Services:
Grant Thornton extends the useful life of its current WAN bandwidth.
Northwestern Mutual Life enforces TN3270 response time service-level agreements.
Information Management:
RR Donnelly uses PacketShaper to ensure satisfactory Citrix performance across its WAN.
Standard & Poor’s protects mission-critical Oracle performance.
Service Providers:
Clear Communications uses PacketShaper to offer tiered services to its customers.
Technology:
Autodesk uses PacketShaper to protect SAP and contain Microsoft Exchange traffic.
Lucent Technology protects mission-critical telnet traffic traversing expensive international WAN connections.
13. Slide 13 A Complete Solution
14. Slide 14 Step 1: Classify - What’s Running on My Network? PacketShaper analyzes traffic running across the WAN, inspecting traffic from layer 2 through layer 7. This is distinctly different from traditional switch, router, or switch-router products that, at best, classify traffic based on IP addresses and port.
Traffic analysis must be more intelligent and flexible than simple address and port based analysis. It is common for multiple applications to communicate over the same port. For example, Web, Pointcast, and BackWeb are all configured to communicate over port 80. If you only used port-based analysis, any policy you set would treat unimportant push traffic and business-critical ecommerce traffic identically. Likewise Published Applications running over Citrix ICA use the same port numbers. More granularity is needed to uniquely identify the different traffic types so you can treat them differently.
Additionally, many applications do not use a static, well defined port, but dynamically negotiate port assignments as they establish connections. Examples of these include FTP and H.323 (Voice-over-IP).
PacketShaper is the only solution that provides the intelligence and flexibility needed to manage traffic according to business priorities.
PacketShaper analyzes traffic running across the WAN, inspecting traffic from layer 2 through layer 7. This is distinctly different from traditional switch, router, or switch-router products that, at best, classify traffic based on IP addresses and port.
Traffic analysis must be more intelligent and flexible than simple address and port based analysis. It is common for multiple applications to communicate over the same port. For example, Web, Pointcast, and BackWeb are all configured to communicate over port 80. If you only used port-based analysis, any policy you set would treat unimportant push traffic and business-critical ecommerce traffic identically. Likewise Published Applications running over Citrix ICA use the same port numbers. More granularity is needed to uniquely identify the different traffic types so you can treat them differently.
Additionally, many applications do not use a static, well defined port, but dynamically negotiate port assignments as they establish connections. Examples of these include FTP and H.323 (Voice-over-IP).
PacketShaper is the only solution that provides the intelligence and flexibility needed to manage traffic according to business priorities.
15. Slide 15 Step 1: Classify – Traffic Class Criteria Inbound/Outbound (travel direction)
Protocol family
Service (very diverse, see online list)
Inside/Outside (location of relevant server)
Port(s)
Service Proxy
IP Address, MAC Address, host name, or host list
Subnet Mask
URL (including wildcards)
Further details (criterion) for Citrix-ICA, Oracle-netv2, HTTP 1.1, RTP-I
Diffserv, IP Precedence, COS/TOS Tips:
Should be very similar to the ideas that the class generated.
Reserve the discussion of Inside/Outside until a later slide.Tips:
Should be very similar to the ideas that the class generated.
Reserve the discussion of Inside/Outside until a later slide.
16. Slide 16 Step 1: Classify – Traffic’s INs and OUTs Example:
1. Sports, Inc. accesses the ESPN web site--an HTTP request. The ESPN web server is outside [/Outbound/Outside/HTTP]
2. The ESPN web graphics are sent to Sports, Inc. The ESPN web server is still on the outside of the Sports, Inc. PacketShaper. [/Inbound/Outside/HTTP]
3.The ESPN marketing folks want to view the Sports, Inc. web site--an HTTP request. The Sports, Inc. web server is inside. [/Inbound/Inside/HTTP]
4. The Sports, Inc. web page is transmitted to ESPN. The Sports, Inc. web server is inside. [/Outbound/Inside/HTTP]Example:
1. Sports, Inc. accesses the ESPN web site--an HTTP request. The ESPN web server is outside [/Outbound/Outside/HTTP]
2. The ESPN web graphics are sent to Sports, Inc. The ESPN web server is still on the outside of the Sports, Inc. PacketShaper. [/Inbound/Outside/HTTP]
3.The ESPN marketing folks want to view the Sports, Inc. web site--an HTTP request. The Sports, Inc. web server is inside. [/Inbound/Inside/HTTP]
4. The Sports, Inc. web page is transmitted to ESPN. The Sports, Inc. web server is inside. [/Outbound/Inside/HTTP]
17. Slide 17 Step 1: Classify – Manual Class Creation Tips:
A class definition is a name, a parent, several other settings, and one or more matching rules
When you create a new class, its first matching rule is embedded in the definition
A class’ parent is defined by which class was displayed when you selected New Class
A class’ other settings can be set only by changing default values after the class exists.
Usual sequences:
Alter a Matching Rule or add a Matching Rule to an existing class.
OR
Create a new class, create a Matching Rule, finish up the class definition.
Tips:
A class definition is a name, a parent, several other settings, and one or more matching rules
When you create a new class, its first matching rule is embedded in the definition
A class’ parent is defined by which class was displayed when you selected New Class
A class’ other settings can be set only by changing default values after the class exists.
Usual sequences:
Alter a Matching Rule or add a Matching Rule to an existing class.
OR
Create a new class, create a Matching Rule, finish up the class definition.
18. Slide 18 Classes are made up of matching rules
Classes can have many matching rules
multiple matching rules are OR’d together
1 rule for each of 3 servers.
single matching rules are AND’d together
1 rule that catches traffic from a specific server to a specific client. Step 1: Classify – Matching Rules Tips:
Additional concept to bring up here:
A matching rule doesn’t need to redefine the subset defined by its parent’s matching rule.
For example:
Subnet
FTP
HTTP
The FTP and HTTP classes don’t need to reference the subnet specified by their parent.
Tips:
Additional concept to bring up here:
A matching rule doesn’t need to redefine the subset defined by its parent’s matching rule.
For example:
Subnet
FTP
HTTP
The FTP and HTTP classes don’t need to reference the subnet specified by their parent.
19. Slide 19 Step 1: Classify – More on Matching Rules The definition of the traffic in a class is a matching rule
It’s a collection of values for the criteria we listed
Traffic Discovery defines matching rules for the classes it creates
You define matching rules for the classes you create
Tips:
It’s part of a class definition.
Every class has at least one TSpec.
Explain the left/right columns, since it’s not obvious.
Explain that the URL Path field is for the portion of the URL that’s AFTER the “www.domain.com.” That first part goes into the host name field.Tips:
It’s part of a class definition.
Every class has at least one TSpec.
Explain the left/right columns, since it’s not obvious.
Explain that the URL Path field is for the portion of the URL that’s AFTER the “www.domain.com.” That first part goes into the host name field.
20. Slide 20 Step 1: Classify – Adv. Matching Rules In addition to the basic criteria, such as IP address and port numbers, the following advanced options are available:
Host Lists
Details for Citrix, Oracle, HTTP 1.1 and RTP
Diffserv and IP COS/TOS
21. Slide 21 Step 1: Classify – Host Lists Instead of a single IP address or a range of IP addresses, specify a list of hosts.
Host lists facilitate traffic classification by enabling multiple, non-contiguous hosts to be specified in a single matching rule. This feature takes advantage of information provided by LDAP directory services. You can configure multiple PacketShapers to reference the same global host list.Host lists facilitate traffic classification by enabling multiple, non-contiguous hosts to be specified in a single matching rule. This feature takes advantage of information provided by LDAP directory services. You can configure multiple PacketShapers to reference the same global host list.
22. Slide 22 Citrix-ICA, Oracle-netv2, HTTP 1.1, and RTP-I can be further classified using the Matching Rule Criterion field:
Citrix-ICA: by published application, client name or priority level*
Oracle-netv2: by database name
HTTP 1.1 by DNS name or IP address
RTP-I (real-time protocol for media streaming) by Encoding Name, Media Type ("a" for audio, "v" for video), or Clock Rate (8000, 16000, 44100, 90000) Step 1: Classify – Application Criteria Citrix classification by client name uses the client name configured in the Citrix Remote Application Manager. If the Citrix server is on a different subnet than the Citrix client, the server location must be set in the Remote Application Manager, which enables Citrix classes with published applications to be classified properly.Citrix classification by client name uses the client name configured in the Citrix Remote Application Manager. If the Citrix server is on a different subnet than the Citrix client, the server location must be set in the Remote Application Manager, which enables Citrix classes with published applications to be classified properly.
23. Slide 23 Step 1: Classify – Diffserv, COS/TOS … Diffserv Code Point (DSCP) (6-bit field)
Value of 0-63
COS - Class of Service (3-bit field)
IP precedence value 0-7
TOS - Type of Service (4-bit field)
802.1q/ISL VLANs
MPLS Applications set the COS field to tell routers how to prioritize packets. For example, weighted fair queuing (WFQ) algorithms in routers use this information.
You can tell PacketShaper to match on these precedence bits during classification (IP protocols only). Then, you can apply a specific policy to manage this traffic type. For example, you could apply a policy that substitutes a different precedence value so that you can control the packet’s priority when it reaches the router.Applications set the COS field to tell routers how to prioritize packets. For example, weighted fair queuing (WFQ) algorithms in routers use this information.
You can tell PacketShaper to match on these precedence bits during classification (IP protocols only). Then, you can apply a specific policy to manage this traffic type. For example, you could apply a policy that substitutes a different precedence value so that you can control the packet’s priority when it reaches the router.
24. Slide 24 Step 1: Classify - Other Settings Tips:
Most of a class’ data fields are in its Tspec. A few are not.
Cover Top Talkers and Top Listeners, Traffic Discovery within Class, and Comment. Exception classes are on next slide.
Top talkers and top listeners track heaviest users (senders and recipients).
Traffic Discovery in class: You might want to have it off for the traffic tree in general, but on for a couple of classes (in a subnet class, for example). (Leave On on Setup screen, turn off for top-level classes (in, out) and on for class.
Ask: Why would you not want Traffic Discovery on?
(You worked hard on customizing your tree and don’t want it altered)
Ask: Why would you want Discovery on for one class?
Ask: When would it not be possible to turn on Traffic Discovery for one class?
(When the class is already the finest granularity of discovery -- HTTP, for example. You wouldn’t want PS to go wild creating classes for every URL or every address.
Additionally, if the class is in/outside, you can’t turn on discovery. The first thing discovery does is create in/outside. Since recursive embedding makes no sense, not allowed.)
Tips:
Most of a class’ data fields are in its Tspec. A few are not.
Cover Top Talkers and Top Listeners, Traffic Discovery within Class, and Comment. Exception classes are on next slide.
Top talkers and top listeners track heaviest users (senders and recipients).
Traffic Discovery in class: You might want to have it off for the traffic tree in general, but on for a couple of classes (in a subnet class, for example). (Leave On on Setup screen, turn off for top-level classes (in, out) and on for class.
Ask: Why would you not want Traffic Discovery on?
(You worked hard on customizing your tree and don’t want it altered)
Ask: Why would you want Discovery on for one class?
Ask: When would it not be possible to turn on Traffic Discovery for one class?
(When the class is already the finest granularity of discovery -- HTTP, for example. You wouldn’t want PS to go wild creating classes for every URL or every address.
Additionally, if the class is in/outside, you can’t turn on discovery. The first thing discovery does is create in/outside. Since recursive embedding makes no sense, not allowed.)
25. Slide 25 Step 1: Classify – Traversing the Class Tree PacketShaper examines all passing traffic.
- Every flow must be assigned to a class.
It traverses the tree to find the traffic’s correct class.
Traversal starts at the top
If you have a special-case class you want searched first, make it an Exception class.
Example: All PCs in a subnet to be treated the same except one. E.g.: DifferentPC
SubnetA
SubnetB
Tips:
Emphasize that traversal is from top, that’s why exception classes must be at the top.
Assume the role of PacketShaper. You note passing traffic.
Start at the top of the tree.
To the traffic: Are you inbound? Yes? Examine children.
Are you from an outside server? Yes? Examine children.
Are you FTP (assuming FTP is 1st child)? NO? Continue to next child.
Are you HTTP? Yes? Examine children. Etc.
Then explain exception classes again. They’re sort of the equivalent of programming GOTOs.
Tips:
Emphasize that traversal is from top, that’s why exception classes must be at the top.
Assume the role of PacketShaper. You note passing traffic.
Start at the top of the tree.
To the traffic: Are you inbound? Yes? Examine children.
Are you from an outside server? Yes? Examine children.
Are you FTP (assuming FTP is 1st child)? NO? Continue to next child.
Are you HTTP? Yes? Examine children. Etc.
Then explain exception classes again. They’re sort of the equivalent of programming GOTOs.
26. Slide 26 Step 1: Classify – RMC After 24 Hrs 55 Applications AutoDiscovered
6 Peer-to-Peer (circled)
7 Streaming
3 Chat
5 Games
And the usual Internet and network service protocols
27. Slide 27 Step 2: Analyze Understanding which applications are competing for the bandwidth and which users are taking more than their fair share will help you zero in on the trouble spots. PacketShaper’s link utilization and top-10 classes graphs are an integral part of characterizing overall network performance, enabling effective bandwidth allocation policies.
The utilization graph shows the peak and average rate. The average rate doesn’t indicate a problem. The peak is the largest 1-second sample.
In the Network Efficiency graph, 100% efficiency means that there were no retransmissions.Understanding which applications are competing for the bandwidth and which users are taking more than their fair share will help you zero in on the trouble spots. PacketShaper’s link utilization and top-10 classes graphs are an integral part of characterizing overall network performance, enabling effective bandwidth allocation policies.
The utilization graph shows the peak and average rate. The average rate doesn’t indicate a problem. The peak is the largest 1-second sample.
In the Network Efficiency graph, 100% efficiency means that there were no retransmissions.
28. Slide 28 Step 2: Analyze - Top Ten Tab
29. Slide 29 Step 2: Analyze - Context-Specific Reports The report types include:
Utilization
Utilization with Peaks
Application Response Time
Shaping Policy Effectiveness
Top-10 Partitions
Top-10 Classes
Custom (user defines the graphs to be included)
The report types include:
Utilization
Utilization with Peaks
Application Response Time
Shaping Policy Effectiveness
Top-10 Partitions
Top-10 Classes
Custom (user defines the graphs to be included)
30. Slide 30 Step 2: Analyze - Monitor Tab
31. Slide 31 Step 2: Analyze - Top Talkers / Listeners Enable up to 12 top talkers/listeners (total).
Create classes for top users. Look at what’s competing for bandwidth.
If it’s a specific user, you can create a class for that user and then discover the applications in use by that user. Look at what’s competing for bandwidth.
If it’s a specific user, you can create a class for that user and then discover the applications in use by that user.
32. Slide 32 Step 2: Analyze - Response-Time Summary View delay statistics for all measured classes:
33. Slide 33 Step 2: Analyze - Transaction Delay RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
34. Slide 34 Step 2: Analyze – Delay Distribution RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
35. Slide 35 Why measure response time?
Quantify performance.
Identify performance problems.
Develop strategies for bandwidth management, server balancing, and topology upgrades.
Assess results after you’ve made configuration changes. Step 2: Analyze – Response Times PacketShaper's Response-Time Advantages
Other applications can be hard on the network administrator by requiring a lot of work or on the network itself by generating additional traffic.
· NO Application modifications
PacketShaper does not require software wrappers around measured applications or the addition of API calls.
· NO Desktop and server changes
Nothing needs to be loaded on client desktops or on any server.
· NO Artificial traffic overhead and NO data collection overhead
No application requests merely to time their responses. It doesn't issue pings.
· NO Router reconfiguration or topology changes
No changes to router configuration, protocols, or topology. Not a single point of network failure.
· NO Location restrictions
Most useful when positioned at the network edge however, PacketShaper measures performance from anywhere on the network, as long as it sees the traffic it's measuring.
PacketShaper's Response-Time Advantages
Other applications can be hard on the network administrator by requiring a lot of work or on the network itself by generating additional traffic.
· NO Application modifications
PacketShaper does not require software wrappers around measured applications or the addition of API calls.
· NO Desktop and server changes
Nothing needs to be loaded on client desktops or on any server.
· NO Artificial traffic overhead and NO data collection overhead
No application requests merely to time their responses. It doesn't issue pings.
· NO Router reconfiguration or topology changes
No changes to router configuration, protocols, or topology. Not a single point of network failure.
· NO Location restrictions
Most useful when positioned at the network edge however, PacketShaper measures performance from anywhere on the network, as long as it sees the traffic it's measuring.
36. Slide 36 Step 2: Analyze - Measuring Delay Server Delay - # of ms the server uses to process a client’s request after all data received.
Total Delay - # of ms from client’s request to receipt of response.
Network Delay = Total Delay - Server Delay
Round-Trip Time (RTT) is the # of ms for client-server exchange of precisely one packet. RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged.
Determine when a transaction ends:
Looks for a Push Flag from the server and the ACK from the client
Not necessarily the real end of the transaction. Often applications with long transactions insert Push flags throughout the transaction.
In addition to monitoring Push Flags, PacketShaper uses a timer to track transactions and also checks a number of conditions
RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged.
Determine when a transaction ends:
Looks for a Push Flag from the server and the ACK from the client
Not necessarily the real end of the transaction. Often applications with long transactions insert Push flags throughout the transaction.
In addition to monitoring Push Flags, PacketShaper uses a timer to track transactions and also checks a number of conditions
37. Slide 37 Step 2: Analyze - RMC Link Performance Once you know what is running on the network, you will want to measure performance. PacketShaper measures end-to-end response time for discovered and classified traffic. Measurements include network and server delay for TCP connections. This feature is known as Response Time Management (RTM).
From these measurements you can establish a baseline that will enable you to quickly understand when performance deviates from the norm for both network and server delay. Further analysis can be performed by looking at the traffic distribution running on the network.
Once you know what is running on the network, you will want to measure performance. PacketShaper measures end-to-end response time for discovered and classified traffic. Measurements include network and server delay for TCP connections. This feature is known as Response Time Management (RTM).
From these measurements you can establish a baseline that will enable you to quickly understand when performance deviates from the norm for both network and server delay. Further analysis can be performed by looking at the traffic distribution running on the network.
38. Slide 38 Step 2: Analyze – RMC – Top Applications
39. Slide 39 Step 2: Analyze – RMC – HTTP/SSL Response Times
40. Slide 40 Step 3: Control – How Do I Control Performance? Set policies to control performance
Per-flow minimum/maximum bandwidth policies
Per-user minimum/maximum bandwidth policies
Priority-based policies
Admissions Control
Partitions for control of aggregate flows
PacketShaper implements TCP Rate Control
Control when and how much data end-systems transmit
Using industry-standard TCP/IP
Manage traffic flows and aggregate classes with bits-per-second accuracy By setting policies, the network manager can guarantee a minimum rate to each flow of a traffic class, allow each traffic flow to use excess rate, set a limit on how much of the excess rate can be used, and keep greedy traffic in line.
For example, a network manager can establish policies that set a bits-per-second minimum rate for a connection. Rate policies smooth bursty traffic and can reserve a guaranteed rate for a traffic class. Rate-based policies are required to protect certain applications like Citrix and Oracle, which require predictable access to a minimum amount of bandwidth for satisfactory performance.
TCP Rate control is a method for preventing congestion from affecting network links by slowing the rate at which traffic is sent, effectively controlling the rate at which systems communicate. By acting proactively and preventing congestion from occurring, TCP Rate control significantly reduces queuing-induced latency and packet loss.
In addition, TCP Rate Control is able to control both inbound and outbound traffic. This is something deficient in queuing-based solutions, such as routers, and is unique to PacketShaper.By setting policies, the network manager can guarantee a minimum rate to each flow of a traffic class, allow each traffic flow to use excess rate, set a limit on how much of the excess rate can be used, and keep greedy traffic in line.
For example, a network manager can establish policies that set a bits-per-second minimum rate for a connection. Rate policies smooth bursty traffic and can reserve a guaranteed rate for a traffic class. Rate-based policies are required to protect certain applications like Citrix and Oracle, which require predictable access to a minimum amount of bandwidth for satisfactory performance.
TCP Rate control is a method for preventing congestion from affecting network links by slowing the rate at which traffic is sent, effectively controlling the rate at which systems communicate. By acting proactively and preventing congestion from occurring, TCP Rate control significantly reduces queuing-induced latency and packet loss.
In addition, TCP Rate Control is able to control both inbound and outbound traffic. This is something deficient in queuing-based solutions, such as routers, and is unique to PacketShaper.
41. Slide 41 Step 3: Control – Applying Policies
42. Slide 42 Step 3: Control – Priority Policies
43. Slide 43 Step 3: Control – Priority Policy Guidelines Use a priority policy:
When rate is not your primary objective
If traffic does not burst (surge)
If traffic is latency-sensitive
If high-priority flows are small, orif low-priority flows are large but not bursty
Priority policies are appropriate for interactive traffic like TN3270 or Telnet (latency-sensitive, don’t burst, small) Tips:
Examples:
SNA over frame relay (can’t do rate control for non-IP anyway).
If HTTP is divided into a text class and a graphics class, then you could use a priority policy for the text (it’s small) and do rate-control for the larger graphics class.
Tips:
Examples:
SNA over frame relay (can’t do rate control for non-IP anyway).
If HTTP is divided into a text class and a graphics class, then you could use a priority policy for the text (it’s small) and do rate-control for the larger graphics class.
44. Slide 44 Step 3: Control – Rate Policy Page Tips:
Cover Guaranteed, Burstable at Priority, and Limit.
Leave Scaling, Admission Control, IP Precedence, and Failover for later.
You are guaranteed a smooth connection at a predictable rate even during heavy traffic.
Ask students for an example of when a guaranteed rate would be needed (Voice, streaming video)
Warning: Since most Web browsers open multiple simultaneous connections, a Web policy set to 100Kbps may acdtually allow as much as 400Kbps per workstation.
Tips:
Cover Guaranteed, Burstable at Priority, and Limit.
Leave Scaling, Admission Control, IP Precedence, and Failover for later.
You are guaranteed a smooth connection at a predictable rate even during heavy traffic.
Ask students for an example of when a guaranteed rate would be needed (Voice, streaming video)
Warning: Since most Web browsers open multiple simultaneous connections, a Web policy set to 100Kbps may acdtually allow as much as 400Kbps per workstation.
45. Slide 45 Step 3: Control – Rate Policy Guidelines Guarantee each flow a minimum bits-per-second rate
Give each flow prioritized access to excess bandwidth
Keep a lid on surging, bandwidth-hungry flows
Guard mission-critical flows
Give delay-sensitive flows a chance
Make sure behind-the-scenes TCP Rate Control is active
Remember not to over-commit guaranteed rates!
46. Slide 46 Step 3: Control – Never-Admit Policies Use a Never-Admit policy:
For TCP or Web traffic, to block a session and inform the user
Tips:
You may want to block a service or application because:
it consumes too much of your network bandwidth
it is non-essential to your business
it is not permitted on your network
For example, you could take a firewall-approach where you do not inform hackers that they cannot access your network, you just leave them hanging.
A Never-Admit policy offers you the choice of refuse or redirect. Redirect requires that you supply a complete URL. Refuse puts up a message that your Web request was blocked, “brought to you by Packeteer PacketShaper.”
(No need to mention Admission Control to students here, Never-Admit UI is self-contained and easily understood.)
NOTE: If you use the discard policy for TCP traffic, the discarded packets will cause TCP time-outs resulting in a long wait period before the user gets any feedback. This may or may not be desirable. For example, it may be desirable to “time out” someone attempting to crack your Web server.
Tips:
You may want to block a service or application because:
it consumes too much of your network bandwidth
it is non-essential to your business
it is not permitted on your network
For example, you could take a firewall-approach where you do not inform hackers that they cannot access your network, you just leave them hanging.
A Never-Admit policy offers you the choice of refuse or redirect. Redirect requires that you supply a complete URL. Refuse puts up a message that your Web request was blocked, “brought to you by Packeteer PacketShaper.”
(No need to mention Admission Control to students here, Never-Admit UI is self-contained and easily understood.)
NOTE: If you use the discard policy for TCP traffic, the discarded packets will cause TCP time-outs resulting in a long wait period before the user gets any feedback. This may or may not be desirable. For example, it may be desirable to “time out” someone attempting to crack your Web server.
47. Slide 47 Step 3: Control – Discard Policies
48. Slide 48 Step 3: Control – Ignore Policies
Ignore policies:
Treat traffic as pass-through
Exempt a traffic class from bandwidth management
PacketShaper does not count the statistics Tips:
Ask students when they might want to use and Ignore policy.
Example:
If the PS sees traffic that is never going to land at the router, it’s destined for, perhaps, a web server handling the intranet. PS passes that traffic straight through, not counting it against the router’s link capacity.
Tips:
Ask students when they might want to use and Ignore policy.
Example:
If the PS sees traffic that is never going to land at the router, it’s destined for, perhaps, a web server handling the intranet. PS passes that traffic straight through, not counting it against the router’s link capacity.
49. Slide 49 Step 3: Control – How flows Compete for Excess
50. Slide 50 Step 3: Control –How Flows Compete For Demand
51. Slide 51 Step 3: Control - TCP Rate Control
52. Slide 52 Step 3: Control – Multiplexing Gains
53. Slide 53 Step 3: Control – Queuing versus Rate Control
54. Slide 54 Deadline scheduling mechanism:
Provides rate control for UDP
Not as good as TCP rate control
Uses a delay bound to
Set the maximum delay
Limit buffer utilization per flow
Allows setting the delay bound from 200 to 10,000 milliseconds Step 3: Control – UDP Delay Bound
55. Slide 55 Step 3: Control – Partitions
56. Slide 56 Step 3: Control –Partitions’ Two Purposes The FTP traffic we just talked about was an example of LIMITING traffic--creating a pipe for FTP traffic that still enabled downloads but didn’t allow FTP to interfere with other traffic.
What are some other traffic types you might want to limit?
Doom
HTTP
Partitions can also PROTECT mission-critical traffic. Say you have customers….you can create a partition for Citrix traffic so that you customers are always going to get x amount of bandwidth, guaranteed.
The FTP traffic we just talked about was an example of LIMITING traffic--creating a pipe for FTP traffic that still enabled downloads but didn’t allow FTP to interfere with other traffic.
What are some other traffic types you might want to limit?
Doom
HTTP
Partitions can also PROTECT mission-critical traffic. Say you have customers….you can create a partition for Citrix traffic so that you customers are always going to get x amount of bandwidth, guaranteed.
57. Slide 57 Step 3: Control – Partitions Can Burst You can:
Create a static partition
Create a partition that can grow (burst) if extra bandwidth is available
Partitions can burst to use:
The entire link
A predetermined maximum amount of bandwidth Partitions can burst--which means that if you set up a partition
You can create partitions that meet your specific needs.
Create a partition that will never be any larger than the size you’ve determined--for example, to discourage web-surfing, maybe you never want HTTP to exceed 20K.
Create a partition that can burst when bandwidth is available--for example, allowing FTP downloads to use additional bandwidth when available.
Partitions can burst--which means that if you set up a partition
You can create partitions that meet your specific needs.
Create a partition that will never be any larger than the size you’ve determined--for example, to discourage web-surfing, maybe you never want HTTP to exceed 20K.
Create a partition that can burst when bandwidth is available--for example, allowing FTP downloads to use additional bandwidth when available.
58. Slide 58 Step 3: Control – Dynamic Partitions Automatically setup and tear down partitions based on active users.
Limit each user to a maximum amount of b/w at all times.
Set a cap on number of active users assigned a partition.
Create an overflow partition for everyone else
Dynamic Partition usage graph
* You can now place a limit of bandwidth per user.
* Setting a cap is an optional feature.
* Creating an overflow partition is an optional feature.
* The Dynamic Partition usage graph shows the number of active users in a dynamic partition and the number of subpartitions PacketShaper attempted to create after the partition’s cap was reached.* You can now place a limit of bandwidth per user.
* Setting a cap is an optional feature.
* Creating an overflow partition is an optional feature.
* The Dynamic Partition usage graph shows the number of active users in a dynamic partition and the number of subpartitions PacketShaper attempted to create after the partition’s cap was reached.
59. Slide 59 Step 3: Control – Creating a Partition Tips:
Leaving the size field empty means there is no minimum. PacketShaper calls this an “uncommitted” partition size on the Monitor screen.
Ask what leaving Limit blank means.
What are the purposes behind the following choices for size, burstable, limit:
1. Blank, burstable, low max (contain, limit, no promises)
2. Explicit size, not burstable, blank (protect and limit)
3. Explicit size, burstable, blank (protect, access to more, important traffic)
Highlight Partitions Summary.
Tips:
Leaving the size field empty means there is no minimum. PacketShaper calls this an “uncommitted” partition size on the Monitor screen.
Ask what leaving Limit blank means.
What are the purposes behind the following choices for size, burstable, limit:
1. Blank, burstable, low max (contain, limit, no promises)
2. Explicit size, not burstable, blank (protect and limit)
3. Explicit size, burstable, blank (protect, access to more, important traffic)
Highlight Partitions Summary.
60. Slide 60 Step 3: Control – Dynamic Sub-Partitions
61. Slide 61 Step 3: Control – Dynamic Sub-Partition Details
62. Slide 62 Step 3: Control – Time of Day Scheduling CLI only
Syntax: schedule <time rge> <cmd> | <-f cmd file>
Use “schedule show” to see scheduled items.
Use “schedule delete <#>” to remove scheduled items.
Schedule commands are stored in RAM so they do not span resets.
To span resets create a file named startup.cmd in 9.256/
Put schedule commands in startup.cmd to change shaping by time of day.
When PS boots up it reads startup.cmd and schedules commands.
To immediately apply a new schedule command delete old scheduled times and enter “run startup.cmd” to initialize the new commands.
63. Slide 63 Step 3: Control – Organizing the Traffic Tree
64. Slide 64 Step 3: Controlling VoIP and Video Traffic
65. Slide 65 Step 4: Report - How Do I Measure Performance and Plan for the Future? PacketShaper lets you make more intelligent decisions
Evaluate the effectiveness of shaping
See what traffic you are spending your WAN Budget on
Plan for the future of your network through capacity planning, trend analysis, etc
Track application service level agreements based on total delay, server delay and network delay
Set and meet user expectations
Import data into other reporting systems
CSV, SNMP, XML
Complex plugins for HP Openview, Concord eHealth, InfoVista, NetCool and other NMS…
Can notify via email or SNMP trap when performance is poor or when there is a possible DoS attack
Once your customers have implemented PacketShaper to discover, measure, and control the applications running on their network, they can confidently set expectations (Service Level Agreements) for application performance to the end users.
PacketShaper enables enforceable service levels for critical application performance across the WAN by protecting your customers’ most important applications from other bandwidth-hungry applications, print jobs and file transfers, and delivering precise bandwidth consumption policies.
Once your customers have implemented PacketShaper to discover, measure, and control the applications running on their network, they can confidently set expectations (Service Level Agreements) for application performance to the end users.
PacketShaper enables enforceable service levels for critical application performance across the WAN by protecting your customers’ most important applications from other bandwidth-hungry applications, print jobs and file transfers, and delivering precise bandwidth consumption policies.
66. Slide 66 Step 4: Report – Establish Acceptable Performance Set a threshold to define “good service.”
67. Slide 67 Step 4: Report – What’s Good, What’s Bad? Thresholds let you easily quantify good/bad service. Get to the Statistics:Response Time window from the Response-Time Summary window. Just click on the traffic class in the Response-Time Summary window.Get to the Statistics:Response Time window from the Response-Time Summary window. Just click on the traffic class in the Response-Time Summary window.
68. Slide 68 Step 4: Report – Monitoring SLAs RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
69. Slide 69 Step 4: Report – RTM: Transaction Delay RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
RTT - An indicator of transit delay that is independent of data size. RTT is the average number of milliseconds spent in transit when a client and server exchange the SYN and its corresponding acknowledgment.
A transaction involving a large amount of data requires the data to be divided into multiple packets. Whereas a transaction’s network delay reflects the total transit time for all required packets, the RTT reflects the time for a single small packet to make its way from client to server and another packet to reverse that journey.
You can use the RTT to determine if a large network delay is due to large transactions or a slow network. If the RTT is much smaller than the network delay, then the transactions were large. If the two averages are close, then a sluggish network prompted the longer network delays.
Average RTT figures are tracked on a per-host basis. When a traffic class’ transactions reference many hosts, the separate figures are averaged
70. Slide 70 Step 4: Report – Worst Clients/Servers
71. Slide 71 Step 4: Report – Statistics Data Dump Extract lists of variable values for any class. Two months of data stored.
Specify:
One or more variables (definite variety)
Time period
Sort order
Individual statistics or sum totals Tips:
Select a traffic class and click Data under Statistics
Specify settings. List of measurement variables is in the manual.
Can use the data from the PolicyConsole, or use your Web browser to write it out in HTML or comma-delimited format.
Will explore in more detail in the lab.Tips:
Select a traffic class and click Data under Statistics
Specify settings. List of measurement variables is in the manual.
Can use the data from the PolicyConsole, or use your Web browser to write it out in HTML or comma-delimited format.
Will explore in more detail in the lab.
72. Slide 72 Step 4: Report – How to Get the Data
73. Slide 73 Step 4: Report – PacketShaper Events PacketShaper Events notify you when thresholds are exceeded.
Currently command line only
Viewable via the Events Monitor
Several steps to set it up.
74. Slide 74 Step 4: Report – Event set up. 4 steps to Event Notification
Identify the mail server PacketShaper will use to send messages.
Identify the recipients of the email
Identify the SNMP Server PacketShaper will send traps to.
Register the event.
75. Slide 75 Step 4: Report – Setting up email notification
76. Slide 76 Step 4: Report – Setting the Recipients List
77. Slide 77 Step 4: Report – Setting SNMP Server
78. Slide 78 Step 4: Report – Defining Events
79. Slide 79 Step 4: Report – Defining Events
80. Slide 80 Step 4: Report – Registering Events
81. Slide 81 Step 4: Report – Event Summaries
82. Slide 82 Step 4: Report – Event Monitor
83. Slide 83 Standard MIBS
MIB II
10 Basic Groups (system, interfaces, at, ip, icmp,tcp,udp,egp,transmission,snmp)
Private MIBS
Packeteer MIB
Packeteer RTM MIB
Step 4: Report –SNMP MIBs
84. Slide 84 Step 4: Report - “Roll Your Own” Reports Useful API’s
PolicyConsole – HTTP/Javascript
XML
PacketWise Server-side Tags
CGI API
85. Slide 85 Report- Custom Reports via SNMP Authentication Step 1: Determine report type
Step 2: Get an example URL from the WUI
Step 3: Replace the respective variables with your new variables
Step 4: Turn on snmp look authentication:
CLI: sys set dataRetrievalUseSMMPPassword 1
Append &SNMPPASSWORD=<community> to end of URL
Step 5: Put new URL in a web page and the graph will be created
86. Slide 86 Step 4: Report – RMC Link Performance
87. Slide 87 Step 4: Report – RMC Top 10 These are screen shots from Epispocal High School.These are screen shots from Epispocal High School.
88. Slide 88 Step 4: Report – RMC Main Apps
89. Slide 89 Step 4: Report – RMC HTTP Response Times
90. Slide 90 Packeteer’s PacketShaper Provides the application infrastructure that enables you to:
Know what’s on your network
Get visibility into and control over bandwidth usage
Control recreational traffic
Reserve bandwidth for teaching, learning, and research
Make intelligent decisions about capacity planning
And much more…
91. Slide 91 Tools http://support.packeteer.com
PolicyConsole API (ask support for it)
Boilerplate Reporting Portal
Stanford PacketShaper email list
Send email to: majordomo@lists.stanford.edu
Msg body (no subj): subscribe packeteer-edu
Archive: http://www.stanford.edu/group/networking/netlists
Initial Shaping Script
Tons of Perl Scripts
Online White Papers
PacketGuide (v5.2+)
FREE Online Training every Friday
Regional Training Classes
92. Slide 92 Questions & Contact Info Questions?
Sean Applegate, Packeteer Mid-Atlantic SE
(540) 972-8711
sapplegate@packeteer.com
Resellers
Stratacache – 937-224-0485
Vector – 513-786-6618
DPS – 513-489-4200
DDS – 216-676-1760