Ethernet won the LAN wars of the 1990s and is gaining momentum in the WAN. Here, I write about acceleration, services and management.
|
|
|
Written by Peter Brockmann
|
|
| |
 The quality of network service is increasingly a big concern for many users and for many applications. That's because more and more people are being exposed to real-time applications running over IP. Sure, some of these are store and forward apps - Xbox-based Netflicks or Zune streaming - and some are real-time - Xbox-based live game playing, VoIP and video conferencing.
Either way, the user's experience is negatively impacted when there is a network outage or congestion. My own Xbox-Netflicks experience has been degraded in recent weeks for no apparent reason. First, I thought it was my physical LAN connection, so I made a new Ethernet connector.
Then, I thought it was my DNS configuration, so I posted my own DNS server and my cableco's DNS server addresses.
Now, I think it's just congestion on the cableco network at a certain time of day, near 4 pm where there may be many kids in the neighborhood playing Xbox live... I'm keeping a log of when the service interrupts so I can speak intelligently about the problem and maybe even see patterns over time.
But, this Internet Health Report tool from Keynote Systems provides a simple mechanism to view inter-network connection quality. The tool presents the matrix of latency results measured in ms, and averaged over 1 hour, four hours or 24 hours. Similarly, I can click on the metric drop down box and choose to see network availability in % over the past 1 hour, four hours or 24 hours. Or, packet loss over the same period.
Users can click on any one origin-destination pair and see all three results for that combination.
This kind of independent, third-party service performance tool will increasingly be useful as users ask the question, what's wrong with my ____ and what can I do, if anything, to fix it? Users will use it to choose carriers or at least integrate it into their troubleshooting practices. |
|
|
Written by Peter Brockmann
|
|
| |
 In a post-IT Expo briefing, I spent a few minutes with Greg Hough, the CTO of GlobalCapacity, who gave me an update of the company's integration with Vanco, a network integrator and newest contributor to the telecom access network pricing/integration/logistics services company. Greg explained that the network integration services business is a terrific complement to the company's Saas and strategic sourcing operations and integration has been very fast. The network integration is an enabler of other business models and strengthens the company's competitive differentiation vis-a-vis other Saas or virtual network operations company by providing a complete logistics operation.
Pendulum Swings to Enterprise WANs?
I have blogged about GlobalCapacity in recent months and in this update briefing, Greg gave me fresh insights about the nature of competition and how the access network is still the biggest bottleneck to effective enterprise network operations. In the past few years, large enterprise network departments have been largely frozen by fear. Fear of change, fear of commitment and fear of changing technology.
Greg's insight is that the current stressful economic conditions have forced companies to overcome that fear and rethink the assumptions in the way they do things. For telecom engineers and managers this means that they have not only worked to rethink their enterprise network and operations, but they have acted and are acting on those new approaches to network design and operations.
Greg argues that the pendulum that had swung in favor of outsourced WANs through virtual services such as MPLS and frame relay are swinging back to enterprise-owned MPLS networks and operations. Customers have CCIEs on staff programming remote router backup processes, but are otherwise underutilized. They have the scale to benefit from lower cost services, 7x24 monitoring and faster responsiveness. GlobalCapacity's unique knowledge and resources provide the skills to automate and coordinate contracts with local access providers around the world. |
|
|
Written by Peter Brockmann
|
|
| |
 While at IT Expo, I met with Ron Thomas, CEO and Marc Goodman, Director of Marketing for Ecessa (pronounced E-ses-sa), a Plymouth MN appliance manufacturer with an inexpensive load balancing solution to enable 'last mile survivability.'
As I listened to Ron tell me about the company's transformation from a manufacturer of DSU/CSUs years ago to a manufacturer of PowerLink and ShieldLink appliances for managing various applications for access line survivability - including SIP sessions, I was reminded of a particularly painful experience one friend had with enabling automatic balancing between two access lines (last mile) from different carriers/service providers. My buddy Sri had one DSL circuit with a certain IP address range coming from SBC (it was a few years ago in a big southwestern state) and a cableco connection with a certain other IP address range coming from his local cableco.
The problem was to assure that users anywhere on the Internet could access his servers regardless of the state of any one access line. I searched Google for possible solutions and so on... it was a sad day in Mudsville when I had to admit that I didn't know...
 ... Ron explained to me that DNS is a hierarchical Internet service where ISP's DNS servers accept their addressing data from other, authoritative servers. In Sri's case, the appliance sits in the enterprise network between the two links. It authenticates itself to the DNS services of an ISP as the definitive, authoritative DNS server for Sri's domain. The server sets a time-to-live feature of DNS records at something like thirty seconds and then round-robins the DNS requests between the two IP addresses, and two access lines. This way both services are exercised and you don't get into circumstances were the backup line was out of service, having failed a month ago, but because it had never been used no-one knew it was out of service until it was needed, and the switchover is less than 30 seconds.
In Sri's case, he used one until an outage and then used the other through a manual process of updating DNS servers, waiting for their propagation to occur and so on. Pain.
PowerLinks can be paired back-to-back to enable bonding between multiple access links and up-to-six remote sites for higher performance, or by using the ShieldLink appliance, higher security including VPN Gateway functionality. The portfolio of appliances can also monitor QoS, allocating bandwidth and switching services onto specific links when jitter, delay and packet loss exceed some administrator-set threshold.
The ClariLink WAN Optimization Controller is able to fortify availability of SIP communications in remote offices. Since the appliance incorporates a SIP proxy and registrar, it is intimately involved in the signaling of all outbound and incoming SIP communications. So, if one of the last mile access links fail during a voice or video communication, the SIP proxy can instantly order the RTP stream of traffic to the alternative link saving the communications path. The demo I heard at the show involved a momentary silence of duration probably less than 500 milliseconds, which is barely perceptible by most users.
With ClariLink users avoid having to redial the call and are able to continue the conversation as if nothing had happened. Others can continue after an interruption of one or more minutes depending on how fast the last mile switchover might have occurred. If this circumstance was at my friend Sri's network, it might have been an hour or more. Clearly, Ecessa shows how survivability pays for small businesses. |
|
|
|
Written by Peter Brockmann
|
|
| |
 Dr Vinton Cerf is at it again.
This time, it's a protocol optimized for interplanetary communications. That's right. Mars needs Internet service, but with a twist.
As reported in The Economist the limited bandwidth, long transmission times and uncertainty about whether the sender and receiver are actually facing each other at the appropriate moment in a transmission set the stage for a multi-stage store and forward protocol.
NASA calls it a delay- or disruption- tolerant network protocol (DTN). The design is pretty simple - routers around the solar system (Internationall Space Station, a Martian satellite, a NASA probe orbiting the Sun roughly 20 million miles from Earth for example) store messages until the next hop closer to the destination is available. Many of the brilliant protocols and services of the IETF now in service over the Internet are appropriate, with minor modifications. Today data transmission between Earth and Mars can take between 3.5 and 20 minutes depending on the relative rotation of the two planets. In future, however this could be consistently closer to the 3.5 minutes than to the 20 minutes as the number of intermediate nodes that push the data closer to its destination appropriately grow.
The article concludes with the expectation that one day, scientists can simple send messages to a Mars rover with the address
This e-mail address is being protected from spambots. You need JavaScript enabled to view it
. Of course, without more safeguards (I would expect some sort of authenticated sender process) that the traffic to the rover is actually relevant and necessary, I'm afraid that the rover's inbox (not to mention the skinny transmission links) will quickly be filled with all manner of invitations for male enhancement cures, earthly credit relief, naughty earth girls and get rich quick schemes!
Otherwise, it won't be very long at all when the first inter-planetary spam occurs. |
|
|
Written by Peter Brockmann
|
|
| |
The front page of Global Capacity's website tells the story as: "We help customers navigate the entire supply chain of global access networks, improving network efficiency, streamlining network operations and reducing network costs."
Although a little wordy, I think the company is really about becoming a trusted expert in access networks. And that is exactly what they have done. Without owning any facilities, they have managed to do business with 350 of the world's largest and most sophisticated carriers, enterprises and state and local governments. GlobalCapacity either sells to, consults with or partners with BT, BT Global Services, T-Systems, Colt,
GMAC,
AT&T, Cisco, IBM and others.
At the heart of this company's unique position in the telecom consulting and services marketplace is its global pricing platform for telecom access services. Formed through the merger of four companies - Magenta net-Logic of Manchester UK provided the pricing software, CenterPath of Waltham MA provided the metro-optical network design, build an dmanage services for healthcare and financial services organizations, Global Capacity of Houston TX brought layer 2 WAN network integration and Virtual Network Operator skills and clients and Vanco Direct (formerly Universal Access) network integration.
Jack Lodge, COO (we both worked at Nortel in the 1998-2001 timeframe. Jack in IT, me in enterprise marketing), pointed out that the last mile market for business - probably $200 billion/year - is primarily characterized as being highly inefficient and not at all price transparent. There are probably 900 suppliers licensed to provide bandwidth around the world and literally thousands of wholesalers, each implementing different strategies for pricing services including zones, distance, citycenter-to-citycenter, so confusion or doing business with whom you know is a big part of the pain that customers experience.
The access network is rarely considered in a strategic sense. Usually it's put together as incremental requirements and incremental contracts are signed and deployed. In modern business, 60-70% of the cost of operating networks is in the access network.
Global Capacity offers three basic products:
- SaaS portal for access to pricing and workflow systems (can place orders directly to carriers) which is listed at $250,000 for setup and $10,000/month for subscription services. Can be customized to enable loading in a carrier's own POPs and rates, and incorporates a Mark-to-Market feature that gives local prices for competitive purposes.
- Consulting practice around network optimization. Hiring independent engineers to think strategically about the network from time to time assures customers that they have the cost-optimized implementation.
- Telecom Expense Management for Carriers (TEM+) includes auditing networks - collecting inventory, invoices and contracted rates reporting on the gaps, and even implementing network updates that reset the cost structure of the service. This is usually a contingency-based price.
With 104 employees, the company plans to move its OTC stock to the NASDAQ in the near future. In the meantime, they'll focus on healthcare network optimizations and ASPs that want/need to incorporate the access network into their service offerings. From a global footprint perspective, Global Capacity has strong North America, Western and Eastern Europe with limited but growing engagements in Latin America, South America and Asia.
|
|
|
Written by Peter Brockmann
|
|
| |
I first met Steve Dukker at Interop in 2007.
Steve Dukker and NComputing the virtualization PC company has shipped their 1,000,000th unit and picked up serious Microsoft-type executive muscle to boot. Most impressive achievement. ChannelWeb reports that Will Poole, formerly of the Windows Client business and the corporate VP in Microsoft's Unlimited Potential Group joins the company as co-chairman.
|
|
|
Written by Peter Brockmann
|
|
| |
 Caltech is the proud academic home of the Jet Propulsion Lab and Palomar Observatory. A private and premier research institution, [[California Institute of Technology]] is also proud of having faculty earn 32 Nobel prizes. Caltech is also the originator of Fastsoft, the CalTech and VC-backed TCP accelerator.
I had a telephone briefing with Dan Henderson, the VP Marketing for Fastsoft, who gave me a low-down on the company.
|
|
Read more...
|
|
|
|
|
<< Start < Prev 1 2 3 4 Next > End >>
|
|
Page 1 of 4 |
