Wednesday, March 21, 2012

QoS (Quality of Service)


On the Internet and in other networks, QoS (Quality of Service) is the idea that transmission rates, error rates, and other characteristics can be measured, improved, and, to some extent, guaranteed in advance. QoS is of particular concern for the continuous transmission of high-bandwidth video and multimedia information. Transmitting this kind of content dependably is difficult in public networks using ordinary "best effort" protocols.

QoS can be improved with traffic shaping techniques such as packet prioritization, application classification and queuing at congestion points. Using the Internet's Resource Reservation Protocol (RSVP), packets passing through a gateway host can be expedited based on policy and reservation criteria arranged in advance. Using ATM, which also lets a company or user preselect a level of quality in terms of service, QoS can be measured and guaranteed in terms of the average delay at a gateway, the variation in delay in a group of cells (cells are 53-byte transmission units), cell losses, and the transmission error rate.

The Common Open Policy Service (COPS) is a protocol that allows router and layer 3 switches to get QoS policy information from the network policy server.

See also: mean opinion score (MOS), VoIP (voice over IP), CoS [class of service]

Class of Service (CoS)


Class of Service (CoS) is a way of managing traffic in a network by grouping similar types of traffic (for example, e-mail, streaming video, voice, large document file transfer) together and treating each type as a class with its own level of service priority. Unlike Quality of Service (QoS) traffic management, Class of Service technologies do not guarantee a level of service in terms of bandwidth and delivery time; they offer a "best-effort." On the other hand, CoS technology is simpler to manage and more scalable as a network grows in structure and traffic volume. One can think of CoS as "coarsely-grained" traffic control and QoS as "finely-grained" traffic control.

There are three main CoS technologies:

  • 802.1p Layer 2 Tagging
  • Type of Service (ToS)
  • Differentiated Services (DiffServ)

802.1p Layer 2 Tagging and CoS make use of three bits in the Layer 2 packet header that can be used to specify priority. Since three bits does not allow for much sophistication in managing traffic, a new protocol, Differentiated Services (DS or DiffServ), has been developed in draft form by an IETF Working Group. Differentiated Services uses a different approach to managing packets than simple priority labeling. It uses an indication of how a given packet is to be forwarded, known as the Per Hop Behavior (PHB). The PHB describes a particular service level in terms of bandwidth, queueing theory, and dropping (discarding the packet) decisions.

The Differentiated Services protocol exists as a set of related working documents from the Internet Engineering Task Force (IETF). [http://www.ietf.org/]

Tuesday, March 6, 2012

Online support: Who sucks and who doesn’t



February 23, 2012
14 Comments


We put Telkom, Vodacom, MTN, Cell C, Virgin Mobile and Neotel’s online support systems to the test
Telkom and Vodacom came out tops in a recent MyBroadband survey which set out to test the online support channels of South Africa’s largest telecoms operators.

MyBroadband sent all the operators a simple request for information on slow broadband speeds, asking them for feedback on how to address the issue. The operators were asked to respond via e-mail (and hence not telephonically).

Here are the results.

Telkom (support e-mail sent on 11:35 on 21 February 2012)
Telkom immediately responded with an automated e-mail message confirming receipt of the e-mail and providing me with a reference number.
At 12:14 (hence only 40 minutes after the initial e-mail was sent) Telkom officially responded to the e-mail and provided the details needed to resolve the issue.

Vodacom (web support form completed 11:23 on 21 February 2012)
Vodacom immediately replied with an automated e-mail with the message “We are looking into your query and one of our consultants will contact you within 24 hours”. The e-mail also contained a reference number.
At 14:57 (3 hours 29 minutes after the initial e-mail) Vodacom officially responded with all the details to solve the problem.

MTN (web support form completed 11:28 on 21 February 2012)
MTN’s online support system does not have an auto-response system, and the company also did not respond to the online support request by the time of publication.

Cell C (web support form completed 11:31 on 21 February 2012)
Cell C responded immediately via an automated e-mail saying “We have received your query. One of our support staff will contact you shortly if necessary.” This is however where it stopped and there was no feedback from Cell C by the time of publication.

Virgin Mobile (web support form completed 11:41 on 21 February 2012)
Virgin Mobile immediately responded via an automated system, saying “Thank you for contacting Virgin Mobile SA Technical Support Team. We are eager to assist you and will be calling you in the next 4 hours.”
There was unfortunately no feedback from Virgin Mobile within the promised 4 hours, and by the time of publication the company still had not responded to the online support request.

Neotel (website was down and therefore no online support was available)
During the time of testing the Neotel website was down. It was therefore not possible to test Neotel’s online support.

*Please note that MyBroadband only tracked e-mail responses (we asked specifically to be contacted via e-mail), and there may have been calls directed at the numbers provided in the emails.
Email support test
Company Auto responder Official response Time to resolve issue
Telkom Yes Yes 40 minutes
Vodacom Yes Yes 3 hours 29 minutes
Cell C Yes No Not solved
Virgin Mobile Yes No Not solved
MTN No No Not solved
Neotel Website down Website down Website down

Thursday, March 1, 2012

Review: Apparent Networks – PathView Cloud

by

In the last few years the world has seen an explosion in services that are transforming IT, like virtualization, VOIP phones, and cloud computing. And, networks have become faster and more critical to the success of the business.

That shift raises a challenge for traditional network monitoring. The focus is moving away from managing servers and workstations that are all on the same LAN. Traditional device-based monitoring does a great job of reporting statistics for servers and network hardware. But, traditional tools usually don’t do a good job monitoring end-to-end performance – especially when part of the network is delivered as a managed service like MPLS. After all, who cares if your mail server is up, if network congestion is so bad that your users can’t get to it? And, if that congestion is on your service-provider’s core network, how would you ever know?

One solution is PathView Cloud, an innovative performance management tool from Apparent Networks. Available since late 2009, it is designed to appeal to small-to-medium enterprise customers.

What Is It?

PathView Cloud is a tool that can monitor end-to-end performance of network paths and connected devices. For instance, you could monitor a connection to a remote office, and PathView Cloud will alert you anytime that path goes offline. It takes things one step further though, and will generate an alert any time the service quality of a path is in trouble – even if connectivity hasn’t been lost.
As the “Cloud” part of the name implies, PathView Cloud is a hosted application. This makes it dead easy to deploy and run, since all the work of configuring and running database and web servers is done for you. The key to gathering measurements on your network is a small device called a MicroAppliance.

The MicroAppliance is included as part of the PathView Cloud license and it does the actual work of measuring network performance. It works by generating a very small stream of traffic, and then it measures how that traffic is returned to determine how well the network is performing. Perhaps best of all, it can do all of this for parts of the network that you don’t even manage yourself – like your site-to-site WAN connections. The traffic it generates is minuscule and won’t cause any network interruptions for users.

The MicroAppliance is shipped pre-configured to grab an IP address on your network and connect back to the PathView Cloud servers. I’ll admit I was sceptical before it arrived. After all, when does anything just “work” without any configuration? But the MicroAppliance turned out to work as advertized. I plugged it in to my LAN, and moments later it was available for configuration via the web-console. Even moving it around to other networks seemed to be no problem. The MicroAppliance worked every time with no additional configuration required.

Configuring paths to test is as simple as entering an IP address or Host name. Once paths are configured, the MicroAppliance will start monitoring the path from the appliance, to the target.
Service Quality Definitions (SQD) can be defined for each path. SQD’s identify conditions that will trigger alerts. As an example in the screenshot below, identifying a device as a File Server will cause alerts to trigger if connectivity is lost, or if latency rises above 50ms. There are a number of pre-defined SQD’s, or you can create your own.

What Does It Do?

PathView Cloud can help manage your network several ways:

Monitoring

PathView Cloud excels at monitoring connectivity. It maintains a dashboard for each path that displays important information like Available Capacity and Latency, as well as number of hops to destination, and performance per-hop.

It can send alerts by email when it detects a problem, and will automatically run a diagnostic report against the path. Administrators can log in to the web-based console to check status and run additional tests if necessary.

Apparent Networks recommends using one MicroAppliance at each of your sites, which enables another feature that traditional monitoring usually lacks: inside-out monitoring, or the ability to test the network from the perspective of your users. Send a MicroAppliance to one of your satellite offices, have a user plug it in, and instantly you can see network performance from a new perspective.


Troubleshooting

Suspect a network problem reaching one of your devices? If it’s not something you’re already monitoring, then try adding it as a path. Diagnostics are automatically run against new paths, and PathView Cloud will provide recommendations to fix any problems it discovers.

Diagnostics are surprisingly accurate. PathView Cloud quickly identified problems like duplex mismatches on network devices. It will indicate the percentage of certainty that a particular problem exists – helping you to focus your troubleshooting efforts on the most likely problems first.

Planning

PathView Cloud is the perfect tool for planning a VOIP rollout. It can perform a voice readiness assessment which will test the network with simulated VOIP calls. It has options to simulate the number of calls, preferred voice codec, and handset or server manufacturer. The report provides a measured MOS score, and a dashboard-style readiness report.

There is also an option to run a data assessment test, which is useful for testing new services to ensure they are performing properly.

Reporting

The Reporting module is basic, but provides a useful dashboard display of overall service quality, and a detailed breakdown of all service violations. A Top Offenders report also helps to identify problem paths that need attention. Reports can be exported to PDF for easy sharing.

The Verdict

PathView Cloud is a great product, but it does have a few small flaws. One limitation arises when testing to some types of target endpoints. For example, testing to a router or switch may not give good results. Busy routers usually drop some ICMP packets, giving priority to normal traffic forwarding . In those cases PathView Cloud may not report true bandwidth numbers, and could incorrectly identify a packet loss problem. A solution to this problem is to plan to target endpoints like servers or workstations.

Another problem arises when testing across a WAN. If you have a 10Mb WAN link, PathView Cloud can only analyze 10Mb of traffic – it can’t expand back out on the other side to test your 100Mb LAN properly. This is a limitation of the WAN more than the tool, and one solution is to install a second
MicroAppliance at your remote site.

But, this does highlight that the tools are really only as good as the operator. PathView Cloud is simple to deploy and operate, but administrators still need a good foundation in troubleshooting networks. In all cases, an administrator needs to look at the results generated and ask themselves “Is this what I expect?” before assuming that a problem exists.

Another minor gripe is that it’s not easy to find the IP address of your MicroAppliance in the web-console. To find the IP, you have to go through several steps to eventually find it in a diagnostic report. When asked about this, Apparent Networks told us that they plan to add this feature in a future release.

A small nice-to-have feature would be the ability to temporarily pause monitoring for specific paths. For instance, suppose you were taking a device offline for maintenance and didn’t want to receive notifications for that path. You could delete the path, but then you’d lose all your historical data. It would be nice to be able to pause, or ignore paths temporarily without losing all of that information.

Finally, once last thing to keep in mind is that PathView Cloud isn’t really a replacement for your traditional monitoring tools. It can’t monitor disk space, it can’t receive SNMP logs, and it can’t measure HTTP application response time. It’s focused on monitoring performance and connectivity.

In spite of this, network performance is becoming so critical to business that it’s well worth adding this tool to your arsenal.

Despite these few faults, we’d still recommend PathView Cloud. It’s easy to configure, easy to use, and very reliable. It also has a wide range of uses, covering everything from monitoring and troubleshooting, to assessment, planning, and reporting. All of this functionality provides great value for the cost of a license.

Pricing is based on the number of MicroAppliances required. Service contracts start at $500 (USD) per year and include one free MicroAppliance, access to the hosted PathView Cloud server software, and the ability to monitor 55 paths.

Check out a Live Demo of PathView Cloud here, or take PathView Cloud for a test drive with a free 30-day trial.

Friday, February 17, 2012

Boosting ADSL in South Africa

by Rudolph Miller - MWEB’s former CEO Rudi Jansen explains what needs to be done to boost ADSL in South Africa

The cost of IPConnect and the price of ADSL line rental need to reduce significantly to boost ADSL services in South Africa. This is according to MWEB’s former CEO Rudi Jansen.

Speaking to MyBroadband shortly after he stepped down as CEO at MWEB, Jansen explained that the biggest costs for ISPs to provide their own services are: local bandwidth for the national network and the cost of the IPC.

“IPC is the biggest headache. If the IPC cost is reduced ISPs will buy more capacity, Telkom will get more money to upgrade the last mile and cost will reduce further,” explained Jansen.

“In addition, the throughput of the last mile needs to be increased in line with usage trends without increasing costs. This is the only way to relieve congested exchanges.”

Sean Nourse, Internet Solutions’ executive for connectivity, echoed Jansen’s views. “The biggest input costs on ADSL Internet Access is the last mile and IPC costs which are all controlled by Telkom,” said Nourse.

“Reductions in these items will have a significant effect on the affordability of the ADSL Internet Access service.”

Jansen further advocated lower ADSL line rental charges (ADSL access costs) which will drive up ADSL uptake in South Africa.

“Lower ADSL prices will entice more people to take ADSL. With 3G connectivity costs coming down all the time, some consumers find it more attractive to take 3G, despite the fact that ADSL is a better broadband technology for bandwidth hungry applications,” said Jansen.

Jansen added that naked ADSL –providing ADSL without the need to pay for a compulsory voice service – will also have a “massive impact on reducing cost of connectivity”.

Nourse agreed with Jansen, saying that better coverage and lower prices are needed to boost ADSL in SA.

“The market has bought into the fact that it is nice to have broadband access to the Internet. We have seen a massive increase in GSM broadband services over the last few years, yet ADSL hasn’t followed similar growth rates. The only prohibiting factors are coverage and price. Telkom needs to address both of these. Once this has been addressed, we should see more uptake in ADSL services,” said Nourse.

Monday, January 2, 2012

Data usage: Smartphones vs dongles « Broadband « MyBroadband Tech and IT News



January 2, 2012
0 Comments



Vodacom provides insight into the data usage patterns and growth on its network.
Over the last twelve months the data consumed by smartphones increased tenfold on the Vodacom network while data used by dongles only showed moderate growth.

This is one of the surprising revelations made by Vodacom CEO Pieter Uys who addressed the media at an event in Sandton.

Uys revealed that the average data consumed by mobile broadband dongles grew from 500MB to 550MB year-on-year. This translates into a 10% yearly growth in data usage on dongles.

Smartphone data usage in comparison grew by 100% over the same period – up from 40MB per month to the current 80MB per month.

Another interesting statistic is that the ratio of dongle/smartphone data usage on the Vodacom network changed from 80/20 last year to the current 60/40.

The reason for this strong smartphone data usage is twofold: there are far more smartphones on the Vodacom network and smartphone users are consuming significantly more data.

Vodacom currently has 1.1 million data dongles on its network, up from 1 million a year ago. In comparison Vodacom has 4.1 million active smartphones on their network.

Uys added that he expects the strong growth in smartphone uptake to continue, especially with the price of Android devices dipping well below R1,000 each.