Apple / Samsung; UK 4G; Cisco, VMWare and others in the Cloud - Telco 2.0 News Review

Posted on

(Ed. Join us next at Digital Arabia in Dubai, 6-7 November, and Digital Asia in Singapore, 3-5 December. The agendas cover the Digital Economy, Digital Commerce and Digital Entertainment in each region.)

It's finally here: OFCOM has made its decision and given EverythingEverywhere the go-ahead to launch LTE in its surplus 1800MHz spectrum. Apparently, they already have test networks in place, so the first UK 4G service could be on line very soon. Until the lawsuits hit, of course, as this doesn't resolve the whole spectrum mess, and is only likely to provoke Vodafone and O2. EE's owners, of course, contend that the "original two" operators already have an unfair advantage because they might get to refarm their 900MHz GSM holdings.

Meanwhile, there was a brief suggestion that two 4G networks were coming, which then vanished. A condition imposed on their merger by the European Union requires EE's owners to dispose of a substantial chunk of the 1800s, and this week they sold 2 15MHz channels to 3UK. There's a detail, though - EE doesn't have to move out until late 2013 for the first 2x10MHz and 2015 for the rest, so 3UK won't be at the LTE start line. However, the proposed rules for the 4G auction are set up to ensure 3UK gets a serve. If they have a chunk of 1800MHz spectrum already, to go with their 2100s, they may well decide to go for the 800MHz digital dividend block, which leaves the Original 2 with the 2600s.

Huawei engineers have been out installing the LTE radios and pulling more fibre to the EE GSM1800 sites, so there's a good chance that some form of service will be available on the 11th of September when the regulatory clearance kicks in. Unless the courts intervene. For its part, 3UK has contracted Samsung for its LTE RAN, which means that it's reducing its commitment to the MBNL network share with (half of) EE.

Meanwhile, the cost of laying backhaul fibre into the countryside may come down.

Telstra gives some details of the next wave of 4G deployments in Australia. A call on the Aussie opposition not to wreck the NBN after more people than expected signed up for 100Mbps service.

At the device end of the tube, Apple won its case against Samsung, successfully defending its patent monopoly on bevelled corners...well, there is some more serious stuff in there, mostly regarding UX interactions, but still, bevelled corners. Samsung gets to pay $1.05bn in damages, and Apple is trying to get a number of phones banned (list here).

Interestingly, that doesn't include the Galaxy S III or the Galaxy Note, which has led some people to question who really won the case. For example, ZDNet argues that a ban on the GS II might lead to more GS III sales, probably nothing but good news. The Monday Note also points out that the hit list covers a small percentage of Samsung's volume and revenue, while GigaOM notes that the fine amounts to two days' revenue and suggests that being publicly outed as producing something very like an iPhone might not be such a bad thing.

In other Android litigation, Google denies it paid any bloggers to take its side, but does release a list of people who commented on the case and had been paid by Google unrelatedly. And Stanford University tries to estimate just how much IPR Intellectual Ventures has bought up.

ZTE hopes to double its sales of smartphones. Nikon launches an Android-based, hybrid smartphone/compact digital camera - it looks remarkably like a Samsung device with a huge lens sticking out of the back we saw back in 2006.

Apple names the new heads of software and hardware engineering. RIM changes personnel in Europe.

A chewy announcement from Cisco - they are making a significant move into software-defined networking and the cloud. In practice, this means that there is now a unified API for programmers working with the whole line of routers, across all three Cisco operating systems, a new virtualised switch product which implements OpenStack standards, and implementations of OpenFlow controllers and agents. The OpenFlow software is only a proof-of-concept, but it's a major milestone that the technology is being taken seriously by the company that makes 80% of the world's routers.

VMWare was having its annual shindig this week and took the opportunity to change a pricing plan which, rather perversely, charged customers more if they managed to run more virtual machines on the same hardware. So, the better use you made of the technology, the more it cost.

The reason for the change, as Ars Technica points out, is that Microsoft's Hyper-V, plus the availability of a free solution in OpenStack, is putting them under price pressure.

VMWare, as a result, is talking up its features and specifically those of its vCloud Director management solution - this High Scalability post argues that cloud deployment is all about automation, which implies they have a point. Product details are here.

In other product news, IBM has announced its new line of mainframes. As the NYT points out, it's a surprisingly big business and one that has a lot in common with the cloud.

Here's some detail about a major Facebook internal project, Prism, which provides a global job routing and balancing service for their Hadoop clusters. Like Hadoop itself (and a lot of Facebook infrastructure), it's an effort to replicate a proprietary Google system. Like Hadoop, Facebook plans to release the code as open-source software. Meanwhile, Google itself has yet another new cloud data-mining innovation, this time a distributed column store (like a key-value store, but the key matches a stream of values not a single value).

Rather than a Chart of the Week, meanwhile, here's a Talk of the Week, from Google's infrastructure king Urs Hoelzle. He says GMail costs less to run than self-hosted e-mail by a factor of 100.

Here's an Amazon Web Services blog post on their high performance computing features in EC2. The post includes a link to launch an EC2 HPC cluster instantly. There's also a HOWTO on running the ArcGIS geoserver in their cloud.

The cloud isn't magic, though: Salesforce's Q2s are pretty dreadful.

China Mobile will store 16GB of your stuff in their cloud.

Facebook's new app for iOS relies for its instant messaging and status updates on a new network protocol invented by IBM Research for M2M applications, interestingly enough. MQTT (Message Queueing Telemetry Transport) is designed to serve devices with minimal power budgets, like M2M sensors, but provide real-time push delivery - a combination which is always challenging for mobile devices.

The Facebook Engineering blog has a detailed technical discussion here.

Elsewhere, dating network Zoosk's CTO tells High Scalability about the realities of delivering instant messaging at scale using XMPP.

Skype has introduced a new photo-sharing feature.

Salesforce's developer blog has an interesting post about using the Tropo API to take calls and transcribe voicemail, then push their content into the Salesforce task queue.

Forcing customers to the Web site doesn't necessarily save on call centre costs, Telstra may have found out.

Virgin Media updates its CPE firmware and breaks the SIP application-layer gateway. Businessweek interviews Viber about their Belarussian developer farm.

Dan Rayburn reports that Akamai has guaranteed AT&T at least $100 million in revenues from their CDN reseller deal.

Juniper Networks, meanwhile, has opted not to build a full CDN solution.

What exactly are value-added services in a CDN?

The BBC pushes a new version of iPlayer, optimises its websites, and does fundamental research on TV user behaviour.

Is TV ripe for disruption?

Mobile infrastructure as catalyst for community solar power. Syria relies on PCCW for transit. When fancy pricing attacks. And Internet advertising's uncanny valley.