Engineer
Sys Admins all around the world will this morning come in to their office to be greeted with a standing ovation by the rest of the business – all of whom will have got in early especially to make some nice fresh coffee, bake some croissants and place their tokens of appreciation in a heap on the desk of the said Sys Admin.
You can picture the scene of joy. The broad grin on the face. The feeling of deepest satisfaction knowing that despite
We are currently seeing an explosive growth in the distribution and delivery of digital video content across both fixed and mobile networks. Four years ago 100 million videos were watched on YouTube every day. It is two billion today. The BBC’s iPlayer launched in December 2007. It now delivers over 120 million requests every day which adds up to 7 petabytes of data a month.
As a result of this, the volume of data carried by mobile operators has risen twentyfold over the last two years (thanks iPhone), and is forecast to grow almost as much again in the next two years. The figures for fixed operators are less dramatic but still very significant.
Regular readers may have noticed the distinct lack of blog posts last week. At least there were none relating to ip networking, the internet, communications or any other subject I normally spout on about.
That’s because I was out of the office most of the week in an exhausting series of meetings and dinners culminating in a couple of days in Lymington on the south coast. I spent the entire time talking about MPLS, Unified Communications, “the cloud” and current regulatory hot topics in the ISP industry.
The picture below shows me coming up on deck with the team for fresh air before moving on to talk about data centres and the future of mobile data / handheld devices.
In case anyone worries we did make sure that we were well fed and watered during the meetings. A supply of premier cru chablis and bordeaux accompanies the smoked salmon/crab/lobster/duck/caviar/aged aberdeen angus fillet steak (delete according to taste).
PS I have a backlog of jobs to get on with this week including some more serious blog posts. Watch this space.
Its tough at the top and it takes a special breed of individual to stay the pace. In this case the pace was being set by Timico Business Partner UES who had sponsored a horse race at the UK’s newest track, Ffos Las near Llanelli.
I have to admire the marketing nouse of UES whose sponsored race was entitled the “Reduce your utility bills with UES Maiden Handicap Stakes”. This message will have been carried in every newspaper in the UK on Tuesday -” The Watchtower” excepted if it is still going.
You may have noticed that the weather was a little damp. It’s what makes God’s Country so green and pleasant.
BT has killed off VoNGA. Bit of a shame really because I was kinda fonda VoNGA. Voice over Next Generation Access or VoNGA was BT Openreach’s initial stab at voice over fibre and initially at least notionally aimed at new developments where it didn’t make sense to put legacy voice infrastructure into an exchange.
Now BT has strangled VoNGA in the womb. We never really heard it’s first cry.
Don’t get me wrong. It was only the acronym I liked – I thought it sounded good. The product itself, a reduced feature
Congratulations to Seb Lahtinen and Thomas Vollrath on their election as non-executive directors to the board of .uk registrar Nominet.
I did an interview with Seb nobbut two days ago and I’m sure that his return to the board will provide value for Nominet.
Nominet has been much in the news this year as the previous government gave itself powers to intervene in the running of the not-for profit org if it felt that the existing Directors were not doing their job.
It is important that we have a steady hand at the helm of the registrar looking after our .uk domain names as businesses become more dependant on the internet.
You can see the Nominet election results here. The system is quite brutal in that your progress as a candidate is highly visible.
Interesting to see that Belgacom ICS , the fourth largest wholesale voice provider in the world, has completely changed over to a VoIP core using Sonus Networks kit. The project, which migrated the equivalent of more than 10,000 E1 TDM and SIP trunks, took only 15 months.
There will be massive cost savings for Belgacom here. Often the business case for this type of migration is justifiable on purely the savings in electricity used to run the old and inefficient TDM switch gear.
In the UK we have the BT 21Century Network but I am not aware that BT has yet undertaken an activity on the same scale as Belgacom. Certainly it has put off migration of exchanges until 2020.
Coincidentally I recently had Sonus in for a chat to look at their new Session Border Controller. Check out the post here.
The IEEE has just ratified the 100GigE standard. We all knew it was coming. It was foretold. I’m not that excited though. I doubt that many people are because 100GigE is currently in the domain of the few. Not many networks have enough traffic to merit using it.
What I am excited about is our own plans to roll out 10GigE across our core network. For me this is a big step forward. Having started an ISP 5 years ago with a 1Gig core it is quite a momentous event. It is being driven by the amount of high speed Ethernet connections being sold to businesses in the UK. There is literally an explosion of them. I have more bandwidth orders in provisioning for the next three months than sold in the previous 5 years.
We are in the UK, I believe, currently seeing a new wave of connectivity in the same way that broadband has grown over the past few years. Broadband has levelled off but in the business world there is a hunger for faster and more reliable connections.
For us the 10Gig investment actually gives us up to 40Gig of capacity and represented a major board level decision. This is not a low cost play. The decison on where to spend valuable capital is sometimes a gamble, albeit one based on careful research. I don’t think this one is such a risk. There is a lot of business out there. You just have to have the right funds to buy into the game.
At work I have a 100Mbps of uncontended bandwidth to play with. It does me. I thought I’d watch some sport this afternoon, in between stuff. This was partly because I drew Chile in the office sweepstake – they are playing Switzerland as I write – and partly because I’m taking one of my lads to Wimbledon on Thursday – centre court – keep an eye out for me in the crowd.
Both sports are being covered on BBC iPlayer this afternoon. I can of course watch both at the same time – and that’s despite being a bloke (or is it because of it?) –
I’ve been tracking the run down of the IPv4 address pool. This morning another two /8s have been allocated and the number remaining has dropped to 6% of the total.
Nov 16 2009 10% – dropped through 400,000,000 mark
Jan 20th 9%
Feb 25th 8%
May 10th 7%
June 2nd 6%
I make no comment here other than it is getting peculiarly exciting. We have effectively used up 5% of the address pool in 7 months.
I woke up this morning to a fault on my broadband line. OK this kind of thing happens. There is a BT engineer coming tomorrow morning to fix it. Fine.
It does however bring home how reliant we are on the internet. I found out in bed at 6.30 am that that the “internet wasn’t working” (after I had been down to make the tea!). My wife’s instant reaction was “OMG what if I get any emails from potential eBay buyers asking questions about my sales items”.
“Don’t worry” I said, “in an emergency you can use my phone”.
I woke up this morning to a fault on my broadband line. OK this kind of thing happens. There is a BT engineer coming tomorrow morning to fix it. Fine.
It does however bring home how reliant we are on the internet. I found out in bed at 6.30 am that that the “internet wasn’t working” (after I had been down to make the tea!). My wife’s instant reaction was “OMG what if I get any emails from potential eBay buyers asking questions about my sales items”.
“Don’t worry” I said, “in an emergency you can use my phone”.
Sometimes when someone is making a statement you just have to sit up and take notice. In this case LINX69 sponsors Prolabs have impressed by giving us hardware.com branded 4GB USB memory sticks. The new bar has been set.
Coincidentally I recently did a screen on all my kids’ USB sticks checking for viruses. They produced 14 of them for scanning ranging from 128MB (virtually useless these days) to a few 500MBs. One memory stick had a virus. Beware.
Anyway if any other vendors want an obective assessment of their marketing freebies you know where to send them:-).
Day1 of LINX69 had networking equipment vendor Brocade giving an interesting talk about the rollout timescales for the 100GigE standard.
The 100Gig standard has taken 5 years from initial ideas to fruition with ratification being expected in June 2010. Coincidentally the 40Gig standard will have taken the same amount of time with a parallel development aimed at the server market.
40Gig kit reuses some 10Gig elements which is what should allow it to fit in the appropriate part of the price/performance curve.
Whilst a number of vendors have announced 100Gig products it remains to be seen how rapidly some of these will be rolled out and adopted. During the last wave of network upgrades (1Gig to 10Gig) many equipment vendors had their fingers burned as industry uptake took a lot longer than anticipated. Nortel, for example had apparently predicted 2 million 10Gig port shipments by 2002 but actually took another 7 years to hit that volume. Somewhat symptomatic of the problems the Nortel business found themselves in methinks.
The upshot is that vendors are unlikely to rush out 100Gig products.
We expect of course that next gen technologies result in lower per port costs. Currently this is not the case for 100Gig due to high optics and component costs. Based on historical trends these are expected to drop in 2011/2012. For the moment 100Gig is therefore very much one for the early adopter.
The chart below shows the timeline between adoption of the standard for each technology. Considering that it takes 5 years to develop a standard and looking at the 2002 dot com bubble bursting date that the 10Gig standard was ratified it is perhaps no surprise that 100Gig was delayed.
An interesting question posed during the Board Election Hustings at LINX69 today was “is LINX getting too big?”
For the uninitiated the London Internet Exchange is a membership owned Internet Exchange where network operators peer with other network operators. This means that they pass traffic between each other free of charge. There is a cost for this – running the “exchange” involves buying and maintaining expensive bits of kit that all members connect to.
This cost however is far lower than the alternative of buying access to internet sites around the world from a commercial supplier – something known as internet transit. LINX membership in theory gives you access to around 70% of all internet routes.
LINX is growing rapidly. The organisation has 357 members with 22 new applications in 2010 to date. Network operators want to join because as LINX grows the benefits also grow.
The question at the hustings is valid though. The problem is that the internet was designed as a robust network able to withstand problems at any given single point. If those networks comprising the internet increasingly connect at a single place then this obviously counter intuitive to the way the internet is meant to work.
Now LINX does operate a very robust network – effectively two networks based on two different vendor equipments. It is becoming an increasingly attractive place to peer.
I can’t tell you what the right answer is. ISPs just need to make sure they have alternatives.
For some strange reason as I travel down to LINX69 in London the on train wifi is directing me to Google’s Swedish website. It would be interesting to find out whether this is down to the service provider network being used by EastCoast Trains or something happening in the Google network itself.
I guess someone at LINX69 would be able to find out.
The Internet Assigned Numbers Authority (IANA) has just allocated regional registries RIPE and APNIC a /8 each this month. For the uninitiated a /8 represents 2 to the power of 24 IP addresses or 167,77,216.
A /8 is the largest block allocation that can be made by IANA and these two have had the effect of bringing forward the x-date, the date for IPv4 exhaustion, by 5 months or so to April 30th 2011.
These blocks are subdivided into smaller subnets for further allocation to ISPs/organisations with smaller requirements such as BT and Timico. Timico has a variety of block allocations ranging from /16 to /20’s.
If you want to know more about IP addressing allocations check out wikipedia. The times they are a changing.
Footnote – a day later the date seems to have bounced back to September – don’t know what happened there. Still not very far off though.
Adrian Kennard of AAISP gave a talk on their implementation of IPv6 at yesterday’s UK Network Operators Forum (UKNOF). Whilst it may not be of huge interest to most readers it is worth taking a look at how the old IPv4 and new IPv6 networked worlds will talk to each other.
The picture below represents a tracert to the bbc.co.uk website. The BBC sits on an IPv4 network. AK is moving AAISP exclusively to IPv6. His customers still need to be able to reach everywhere on the internet and this is done by Network Address Translation (NAT), something that most people will associate with private internal IP addresses.
The tracert clearly shows the long originating IPv6 address 2001:8b0:0:31::51bb:1ffa and the point in the network at which NAT is used to convert to IPv4, in this instance when connecting to the LONAP peering exchange. The shorter 212.58.238.129 address is the more familiar IPv4 format.
Thanks for Adrian for permission to use this. His presentation can be found here.
Coming down the broadband pipe this year to a telephone exchange near you is BT’s Real Time QoS product. QoS might not be of interest to your everyday surfer. It does however have the potential to revolutionise the User Experience when using the internet with better control of voice and gaming quality.
What is QoS?
Quality of Service (QoS) in a network is usually the term used to describe the process whereby certain types of network traffic are prioritised above others. QoS is typically required in a network where time-critical applications are being supported. In most cases this means Voice over IP or video but can be applied to financial transactions and gaming (to improve the experience). There isn’t a definitive list. By and large if you design a network with enough capacity to accommodate your bandwidth needs you don’t have to implement QoS.
In practical terms, where ADSL is concerned, this usually means providing a dedicated broadband connection for
Wholesale Broadband Connect (ie broadband on 21Century Network) was launched by BT two years ago this month. There is always pressure to get new products out in a timely manner and BT is no different though their job is harder because the size of the ship makes for difficult steering.
The first (brave leading edge) service providers started to offer ADSL2+ services based on Wholesale Broadband Connect in the Autumn and by March 09 there were 10k subscribers. At this point BT probably kept quiet about the actual number of customers.
One year later the dial has moved significantly and BT now talks openly about the size of their 21CN estate having broken through the 1 million customer barrier in March 2010. In the intervening year their systems have also improved and fault rates come down to approaching those of the mature 20CN broadband product, ADSL Max.
I realise that we all want every thing faster, better, sooner but progress is being made. Still a long way to go mind…
IPv4 address space is down to 7% on the exhaustion counter – see the right hand column on this blog.
I’m never sure when it clicks over because I’m not watching it all the time. I’d like to be there when it clicks down a number. It’s a bit like seeing the mileometer in your car click over a significant mileage – as my Peugeot did recently when it hit 240,000 :-).
I started watching IPv4 addresses in January 2009, certainly as far as this blog is concerned. The number dropped below 10% in September 2009 so in 6 or so months another 3% has gone. Not long now.
I’m going to organise a party next year to coincide with the notional IPv4 exhaustion date. If you want to come get your name down here. I envisage this will be an international event.
I’m chairing a High Definition Voice Workshop in London on 27th April if anyone is interested in coming along.
The workshop is being organised by the Internet Telephony Service Providers’ Association and will be held at their premises at 111 Buckingham Palace Road, starting at 2.30.
A BT employee has reported through our tech support grapevine that apparently the London Fire Brigade pumped 6 million litres of water from the flooded (and burnt) Burne House Exchange in Paddington yesterday.
Sounds like a self extinguishing fire system to me.
We had a quite a few customers affected by the incident with some still offline at the time of writing.
O2 tell us this morning that a flood has taken out the Burne House Paddington telephone exchange in London – they have 115 cell sites down as a result. Apparently fire engines have been seen pumping water out.
Funnily enough The Register is reporting a similar story but this time the Paddington Exchange is down because of a fire!
I suppose they could both be right – pump in water to put out fire – pump water back out afterwards!
It ain’t April 1st until tomorrow so it must be true.
5pm Wednesday – latest news is that the exchange will be down until midday on 2nd April – you heard it first on trefor.net !
I don’t normally go about overtly selling Timico in this blog but sometimes, when the customer is doing it for you in the trade press, a man’s gotta do what a man’s gotta do*.
Flagship Housing is a recent customer of Timico. We implemented an MPLS network for them. They love it enough to
Paid a visit to Bletchley Park on Saturday with a crowd of friends. For the uninitiated Bletchley Park was the nerve centre of the Allied effort to listen in on enemy communications in World War 2. Most will have heard of the Enigma Machine that the Germans used to encrypt their communications. A number were there on display, under heavy supervision, since one of them was stolen in recent years.
Paid a visit to Google in London on Friday morning. Fascinating offices. Refreshments available in reception whilst you checked yourself in and then breakfast from the breakfast bar afterwards.
I was there to meet Google’s regulatory bod, Luc Delany to discuss the company’s approach to the Digital Economy Bill. Google has a fairly light approach to regulation – the company claims to spend only 10% of what Microsoft spends on lobbying.
The Government, with it’s recent embarrassments, would do well to note that in the USA organisations have to disclose how much they spend on lobbying – down to individual lunches. It would be interesting to see how this mapped out in the UK in respect of the DEB stakeholders.
Of course clause 17 (or is it 18 now?) is the one that concerns Google. That’s the one that potentially gives Peter Mandelson powers to decide which websites are acceptable and which aren’t.
There is a feeling that we have all done everything that can be done on the DEB now, aside from last minute noises. We are now just waiting for judgement day. If the DEB does go ahead in its current form I will say that there is going to be one heck of a stink during and after the election.
I have to say that visiting Google is a pleasure. We played a bit of pool, I picked up an electric guitar and had a strum and spent some time in the surround vision version of Google Earth. It has a name but I can’t remember it. You stand inside a ring of large monitors and move a joystick to guide yourself around the world. Great.
Also chilled out in a deckchair for a bit (in the atrium – under the palm tree) before we both had to move on to other meetings. Life’s a beach.
One thing I didn’t realise is that much of the development of Google mobile operating system, Android, happens right there in the London offices. It made me proud to be a Brit (don’t often say that – I’m normally proud to be Welsh).
Most people won’t have given much thought as to how their email gets from A to B or how that video arrives from YouTube. It just comes down the broadband connection which is plugged into the router next to the phone (or somewhere like that). Right?
Well today the worlds biggest router manufacturer, Cisco, announced their latest and greatest product. It isn’t something that you will want to plug into your phone line though because it would take up most of the living room and there wouldn’t be enough room left for the sofa.
It would also be a bit of an overkill because this router, the CRS-3, is powerful enough to handle up to 322 Terabits1 per second, which is roughly a hundred million times faster than the average UK broadband connection speed!
As the clock continues to tick on the IPv4 exhaustion counter I note that we have dropped down to 8% of addresses remaining. I don’t know when this happened but I certainly get the feeling at as we approach the end it is speeding up – I was expecting it to slow down as people conserve address space.
Anyway I’m pleased to say we have now rolled out IPv6 across the Timico core network. Not open for business yet but we are getting there and we do have a few trialists up and running.
Any bets on when the counter gets to 5%?
VoIP QoS should not be an issue if the network is managed correctly. This means the LAN, the WAN link and the core network of the service provider.
If the WAN link is an ADSL then it can be susceptible to congestion at the exchange though in my experience this very rarely happens, even at times of extreme network usage such as the Olympics or last summer’s cricket. Problems here typically stem from underprovisioning of bandwidth. An Internet Telephony Service Provider should also be operating an uncongested core network and a properly designed LAN should never give problems.
Problems do still occur but if you have the right tools these should be straightforward to detect and sort out. One of the ways we manage our network is by using probes embedded at key point in the network.
The diagram below shows the output of one such probe earlier this morning. We are looking at connection downtime, lost packets or packets arriving in the wrong order, jitter and latency or delay. It can be seen that latency is almost the only measurable effect. All the other numbers are too low to count. Even the worst case latency figure seen here of around 46ms is not going to be noticed by the human ear and most of the calls are below 20ms.
This is a very useful tool for IT managers having to run multiple services over a multi-site Wide Area Network and allows them to spot problems before anyone notices and starts to complain.
