Rather than sending out Christmas cards, we’ve supported a couple of charities this year, you’ll find details below.

Best wishes,

Stu.

The Pepper Foundation (http://www.irhh.org/Pepper/)

The Pepper Foundation is a registered charity, which provides funds for The Pepper Children’s Nurses managed by The Iain Rennie Hospice at Home Service (IRHH). Their nurses provide much needed professional and loving home care on a 24 hour, 365 days of the year – on-call basis, for seriously ill children throughout the Hertfordshire and Buckinghamshire area.

Tiggywinkles (http://www.sttiggywinkles.org.uk/)

Every year in Britain over five million wild animals and birds are injured as a direct result of their encounters with man’s world.

Tiggywinkles treats over 10,000 animal casualties every year. Its doors are open 24 hours a day, 365 days a year to receive British wildlife casualties. They treat all casualties free of charge.

It’s a simple but effective mantra that has stuck with me ever since, as it applies to practically any project or process improvement. If you can’t establish a baseline, if you can’t quantify the here-and-now, how can you expect to demonstrate the effectiveness and benefit of change?

So it is with developing a WAN optimization strategy. Before you implement caching, deduplication or acceleration, it can only make sense to understand what’s happening on the network, by developing a baseline of line utilisation and the existing application mix. With this information, it is then possible to define an approach which attempts to align what the network delivers with what the business needs. WAN performance management is almost as much about controlling the applications you don’t want as it is about traffic acceleration – it’s easy to spend thousands of pounds on appliances which make unwanted traffic fly, whilst doing nothing for those business critical applications.

Controlling network traffic is nothing new. Early pioneers of the technology, such as Packeteer, recognised the value of visibility and control years ago, and their kind of products, along with (to some degree) network Quality of Service, facilitated early IP voice and video services through traffic prioritisation.

Then came the acceleration vendors. “With these boxes, whatever you have we’ll make it faster” they said. And in some extent they did. Caching and deduplication can allow massively more data to traverse the WAN. Unfortunately they offered little control over what was being prioritised, so some file transfers may have been improved, but so was the speed of all the unwanted traffic; the peer-to-peer, the personal hotmail access, the Tesco online shopping*, etc.

Exinda were the first mainstream vendor to appreciate that both the above approaches had merit, but that they should be complimentary, not existing in separate boxes with no co-operation (place all your traffic in a proprietary tunnel, then how can it be viewed or controlled?). They introduced the concept of Unified Performance Management and produced a single appliance that has the visibility and control of a packet shaping device, with the performance benefits of an acceleration controller. And we’re not just talking about “this is web traffic, this is a Citrix app”, Exinda’s layer 7 visibility shows exactly what web traffic, which Citrix application: proper deep application inspection. Coupled with comprehensive real-time and historical reporting, that provides the baseline, the measurement.

That layer 7 technology then facilitates control at the same level of granularity. Some web sites may be business critical, others purely recreational. With layer 7 control, individual URLs can be classified and prioritised separately, while voice calls are guaranteed network access and software updates can be forced to load in the background when no other applications are demanding network access. That’s the management.

The next piece is the acceleration itself. Technologies such as caching and deduplication (wan memory in Exinda-speak), manipulation of the actual tcp, spoofing acknowledgements, aggregating small packets into larger segments, all contribute towards getting more data through the same line, more efficiently. Allied with the advanced visibility and control functionality, it’s possible to ensure that only key applications get the acceleration benefit.

Times change, the industry matures, and other vendors are moving towards their own UPM solutions. Exinda’s head start, however, means that they’re already looking at how next to deliver an improved user experience, with object caching for videos, voice call quality assessment, and integration with enterprise management and reporting services. Their same core principles remain, however: you have to see and understand traffic it before you can optimise it. You can’t manage what you can’t measure.

*Other supermarkets exist.

EDN has been working with Exinda for over five years. We have extensive experience of delivering the benefits of Exinda-based WAN optimisation to organistions of all sizes, from single unit traffic shaping deployments, to corporate global networks. We have delivered solutions over leased lines, internet VPNs, satellites and global MPLS covering Europe, Australasia, Asia and the US. If you want to find out how we and Exinda can help you develop your own WAN optimisation strategy, please get in touch.

The Exinda Edge Cache enables single-sided caching of Internet-based content, including web objects, videos and software updates with a single Exinda WAN Optimization appliance at the branch office or data center, enabling a superior user experience and reducing WAN utilization costs. With the Exinda Edge Cache, web objects are cached at the network edge when they are downloaded from the Internet. These objects can then be delivered to the users on the corporate local area network much faster, providing a better user experience. By caching these web objects in the local office, you can drive down the network traffic consumed by each office, which reduces network costs.

With the ExOS 6.1 firmware release, Exinda has expanded Optimization Scalability, which significantly increases the number of accelerated connections on Exinda’s current hardware platforms, allowing more users to be supported on a single appliance. The result is improved performance for more users, without the need for hardware upgrades.

Support for IPv6

Exinda simultaneously provides all of the capabilities for IPv6 as it does for IPv4 traffic, enabling network managers to see, control, optimize and report on both types of traffic on the WAN while service providers and enterprise clients manage the transition between v4 and v6.

Enhanced reporting

With ExOS 6.1, Exinda has expanded its reporting from Top 10 graphs to include “All Other”, which provides perspective on what percentage the top 10 traffic types represents with respect to the entire traffic load.

Enhanced Network Health Alerting

Exinda has expanded its Application Performance Monitoring and Service Level Agreement (SLA) capabilities, allowing for finer-grained monitoring of application performance and alerts for specific TCP and network health issues.

New and Updated Signatures

Exinda’s UPM solution has the ability to identify and manage more than 2,000 application signatures. With ExOS 6.1, more than 70 application signatures have been added and/or updated.

We may look to using schedules to give the replication high priority access over night and reduced during the day, whilst still allowing it to burst to 100% when the WAN isn’t being used for other activities.

iSCSI (Internet Small Computer System Interface) isn’t currently in the Exinda pre-defined applications list.

The components we identified for iSCSI applications are:

1. iSCSI-target (general communication) on tcp port 3260
2. iSCSI-system on tcp port 860 (note that this is an IANA port allocation, but the default and well-known port should be 3260)
3. iSNS (Internet Storage Name Service) on tcp port 3205

We can define the individual applications via the GUI or the command line, then create an application group to cover it all off in a single policy.

(config) # application iSCSI-target port 3260 protocol tcp
(config) # application iSCSI-system port 3205 protocol tcp
(config) # application iSNS port 860 protocol tcp
(config) # applicationgrp iSCSI application iSCSI-target
(config) # applicationgrp iSCSI application iSCSI-system
(config) # applicationgrp iSCSI application iSNS


Then we can check all looks fine with:

# show applicationgrp iSCSI
Application Group: iSCSI
iSCSI-system
iSCSI-target
iSNS


With this application definition, it’s now possible to set Exinda policies to provide visibility, Quality of Service and acceleration to this kind of traffic.

Reduction Statistics by Application
Application: iSCSI
LAN Data (MB): 171768.62
WAN Data (MB): 51275.43
Reduction Ratio (%): 70.15

Effecting data throughput

During the initial replication we were seeing 90-95% average reduction (initial snapshots contain far more duplicated traffic, as there are big repetitive blocks of empty space, windows system files, etc in the LUNs). Now we’re seeing replication of changes, hence the drop in maximum reduction ratio. It’s still a very impressive run rate and a massive saving on overall WAN load. In addition, we’re still able to operate QoS (Quality of Service) on these circuits, so we are able to handle all this replication traffic whilst protecting the user experience for key business applications. It’s because of this level of visibility and control, in addition to these obvious performance advantages from Exinda acceleration, that we continue to endorse, support and recommend the Exinda appliance as a first-class WAN optimisation controller.

(config) # application "CPS" port 1804 protocol tcp
(config) # application "CPS" port 1804 protocol udp
(config) # application "CPS" port 20481 protocol tcp
(config) # application "CPS" port 20483 protocol tcp
(config) # application "CPS" port 20485 protocol tcp


A simple policy permitting the CPS application to run at all times with low priority and a low minimum guarantee is a great way of gaining continuous replication. In this way, CPS can run all the time and take as much bandwidth as is available when no other applications are using the WAN. No bandwidth is wasted, and the user experience is maintained, independently of the bandwidth restrictions within CPS itself.

When we then add Exinda acceleration, we see pretty good results, in excess of 50% reduction:

Here’s an example of how accelerating CPS traffic can really provide throughput benefits; this screenshot show CPS throughput on a 4Mbps internet circuit:

Exinda is installed and maintained in network deployments that exceed 1,000 devices. Exinda products help over  1,500 organizations in 80 countries worldwide to reduce network operating costs and ensure consistent application performance over the WAN. The Exinda Unified Performance Management (UPM) solution encompasses application visibility, control, optimization and intelligent acceleration – all within a single network appliance that is affordable and easy to manage.

Exinda provides users with the best application experience possible by maximizing the speed and efficiency of the wide area network. Exinda’s unified performance management (UPM) solution brings together a suite of advanced, best-of-breed, visibility, traffic-shaping and WAN optimization technologies into a single, easy to-use appliance designed to improve network performance.

Exinda is the only Unified Performance Management solution on the market. Unified Performance Management provides users with a tightly integrated, cost saving solution to get the most out of your network. All controlled through a unified management console.

The company draws on experienced practitioners with international experience in delivering a diverse range of ICT solutions into industries.

The company draws on international experience in delivering a diverse range of ICT solutions into industries.