Dr. Steven Johnston

Welcome to my homepage. Any comments, corrections or suggestions are more than welcome.
Home
About
Blog
Forums
Activities
Documents
Contact
Site Map
My blog of interesting technology
19 November

The PDC second keynote


Wednesdays keynote at PDC was very exciting, not least as we were all given a netbook, which in-itself looks interesting. Microsoft is not traditionally a hardware manufacturer, but now they have the Zune, Xbox and a laptop, what next?

Here are a few of the highlights, in no particular order.

Internet Explorer 9.0 was mentioned, indicating that there will be a strong focus on standards  e.g. HTML 5 and compliance, although it only scores 32/100 on the Acid3 tests so far. I also discovered that the rendering engine in IE is called Trident, which is not to be confused with the Microsoft Research project for workflow called Trident.

Silverlight 3 has just hit the ground and now we are talking about Silverlight 4!, It looks like there will be some very good changes such as Direct2D. The comparison between GDI and D2D looked amazing, and it is worth noticing that it uses hardware acceleration. (And does not require a code change). There will be the ability to right-click, access local hardware and run as a fully trusted standalone application.

There was a brief section on how feedback helps Microsoft steer the direction of new products, and they showed interesting user studies for the snap and shake features in Win7.

As the ribbon is now standard across many Microsoft applications for the better or the worse, the jury is still out for me. It did highlight an interesting point. If you use a ribbon in you application it will automatically be ready for some of the new input devices such as touch screens (part of the reason we got the free acer netbook)

SJJ



08:10 GMT  |  Read comments(0)

17 November

PDC Keynote: Pinpoint, ‘Dallas’ and AppFabric

 

Well the PDC Keynote was very interesting.

Microsoft Pinpoint was announced, which looks like a market place for cloud (Azure) applications, In addition there was the announcement for a data publication codenamed 'Dallas'.

Dallas looks really amazing, once a dataset is published you can consume using various methods. It automatically provides a c# proxy class and even provides the ability to import the data directly into PowerPivot (Another very cool application). Dallas runs on SQL Azure and used ADO.NET Services to publish the data.
Anyone wishing to publish datasets should seriously consider this!

Another announcement was the AppFabric (Download) which will let you run a scale out fabric on your local data centre. This is the same fabric as will be running in the cloud, making it easier to transition between cloud and on-premises applications.

In the near future Azure will be able to host your own images, which you get full admin rights and RDP access.

There is the ability to select what ‘size’ of machine you would like to run you cloud application on. There were a few categories, but interestingly they go as far as letting you get a 8 cores with 15GB of RAM. So quite a powerful machine.

Finally there was a very interesting announcement around the project codenamed ‘Sydney’ this uses IPV6 + IPSEC to bridge the cloud machine instances into your existing infrastructure! So you can utilise the cloud instances as if they were local machines.

SJJ

12:08 GMT  |  Read comments(0)

14 November

I am off to TechED Europe (EMEA)

As events go, this is ‘event of the year for me’ and I have been lucky enough to secure a ticket to this years event. [TechED]

After a bit of a dry spell in the blogging world I have decided to try and blog about any cool technology that I find at TechED, mainly to act as a reference to remind myself which technologies to pursue, but also as tips for others.

TechED is in Berlin this year from 9-13 November.

 

SJJ



04:16 GMT  |  Read comments(0)

13 May

Axum, yet another programming language

 

Ok so I am not too sure we needed another programming language but this one caught my eye. The way things are going before long there will be a need to program many-core CPU machines. Each core may not be powerful so it is important to utilise a core to its maximum. However writing code to run in parallel is often painful (e.g. MPI/OpenMP) so i have been looking at alternative ways to run code. Some of the distributed LinQ and distributed F# features look promising, but then i cam across Axum.

Axum is kind of like F# in that is is geared towards functional programming, but it has a rather interesting way to exchange messages using a channel, which can act like queues. The ability to add a protocol (a state transition protocol) to a channel helps avoid deadlocks. Having done the first few examples it looks good and it would appear that the ‘infrastructure’ is responsible for spawning the correct number of threads depending on how busy the channel is.

Any serializable object can be put in a channel and there is the ability to have a domain which have properties accessible by worker threads.

For those interested:

Look here: http://msdn.microsoft.com/en-us/devlabs/dd795202.aspx

Read this: http://download.microsoft.com/download/B/D/5/BD51FFB2-C777-43B0-AC24-BDE3C88E231F/Axum%20Programmers%20Guide.pdf

 

Back to main site



00:31 GMT  |  Read comments(0)

08 April

My .tel Update

As promised here is an update to my previous .tel blog.

It was good of Justin to get back to me and answer some of my concerns. (Worth noting that is is a Telnic employee)

I think the long and short is be careful what information you disclose on the .tel (As with any publically accessible data).

The updates are quick to appear and the registration was painless. It did take 24hrs to become visible after registration, (as expected) and then any updates I make to the record are instant.

Telnic and Yellow pages are just companies, would you want a company to have control of everything …..

If you run the .tel through the w3c validator :

This document was successfully checked as XHTML 1.0 Transitional!

Valid XHTML 1.0 Transitional     Valid XHTML 1.0 Transitional

So good news that it complies with standards.

Having had the .tel for a few weeks it is the surprising things that bug me, which mainly revolve around customisation. (Justin mentioned this may be possible in the future)

  • The heading, why on earth would I want the heading to any domain page to be the domain? It hurts so much to see a badly positioned/spelt/punctuated heading.  I would at least want a capital letter (or 2) and a space. Also I hate purple. And that damn logo sucks to much (even if it has the required shadowing to demonstrate which decade it was created).  The icons are a bit boring but acceptable. The main page background is a bit of a strange one, but fine if i must … but purple fonts??? Here is what i am talking about, my heading on my .tel is:

stevenjohnston

  • Some people run analytics on pages to see if they are being used and this is not possible.

 

Back



01:21 GMT  |  Read comments(1)

24 March

The new .tel top level domain
[Today: 24March is the day .tel enters the ‘landrush’]

[Well I managed to secure my .tel domain http://stevenjohnston.tel/ and have started to configure it. I will write an update later]

So there is a new Top Level Domain (TLD), ready to join the plethora of TLD’s that already exist. So you have to ask yourself is this new .tel just another domain that is designed to make me part with my money, and another way for cyber-squatters to extort cash from unsuspecting individuals or companies?

Some of the more recent TLDs that have been released include .eu (Europe) and .ch (China), but .tel really does look that little bit different. The key difference is that .tel does not have a webpage, all the information is stored in the Domain Name System (DNS) records.

How does it work?

So when you enter the address of a webpage i.e. www.google.com your browser does not know where to find that webpage. The world wide web relies on IP addresses to locate the page requested, the IP address is a number like a phone number for a computer. The mapping of web addresses to IP address is done using a Domain Name System (DNS). So every page request has 2 calls, one to resolve the IP address and one to retrieve the webpage.

This has several advantages to ensure there is a reliable services, the DNS is replicated, easily available, small and fast. Every user has multiple DNS servers that are available, if a DNS is unable to resolve an address then it passes the request to a ‘parent’ DNS, thus all DNS form a rather large tree structure reducing single points of failure.

The interesting point is that IP addresses are hard to remember, they usually resolve to a single machine and change with the location of the machine. This is analogous to phone numbers, they usually resolve to a single phone [or house] and when you move home/office you change numbers. I realise that companies let you move your home number but this usually takes time (up to a month!) but what about your ‘office’ number, when you change job your office number changes.  So what is required is a fast mechanism, to update numbers so that your ‘name’ resolves to a number (often unknowingly to the user) which then establishes a communication the same way DNS does for web addresses…. well enter .tel.

The full .tel documentation can be found here and is worth a read if you are serious about .tel. I have attempted to list my excitement and concern over the .tel below:

Advantages

  • The entries in the .tel are ready to click-through to an application. So for example my Skype number will launch Skype. It will be interesting to see what devices support this, and what devices get this correct.
  • If you point a browser at the .tel there is an option to download a vCard. There is a caveat in that it may not work on all devices!
  • You can add custom entries (handy if you want to share game handles)
  • Updates are unlimited, so changes can be frequent
  • Some parts can be marked as private. (see disadvantages)
  • There is no need to build/maintain a website.
  • Subfolders can be used to organise data, I can only see this being useful for companies. The subfolders really replace the ‘press one for sales’ at the start of a call, as you could navigate the .tel folder structure then call the correct number correctly.
  • The keywords look like they could be handy to help direct web searches to your page/identity. They can be used to help uniquely identify you from all the other people with the same name across the world.

Disadvantages

  • This looks like a spammers heaven!
  • I wonder how it would assist in identity theft. I suggest that owners of a .tel think very carefully what information they want the entire WORLD to know about them.
  • I am intrigued to know where the ‘data’ will actually live. If you have a domain (not a .tel) and apply changes e.g. change DNS or CNAME it takes up to 24hours (i have even seen 72 hours quoted) for the update to propagate through the DNS servers.  With .tel they ‘claim’ that an update takes a minute, which indicates that the data is stores in a single (or relatively few) DNS servers. So i wonder how well they will cope with billions of users adding/changing the records. Is it a single point of failure and a potential network bottleneck.
  • Telnic are just a company. Is it a good idea to let them control and own a directory for all users and companies in the world?
  • On the outset it looks like .tel is a lightweight directory, but when you consider that it provides users with the ability to have private data, locate friends, group friends, send friend request messages, it all starts to get a bit bloated.
  • The data limitations are small and DNS was not intended as a tool for this kind of data. There may be scalability issues. I could not find any limitations on the number of entries, or subfolders permitted (a subfolder is a a sub domain)
  • A subfolder (sub domain) cannot have private data! This is strange, and makes me wonder how the ‘private’ data is held .
  • When you point a browser at a .tel it is rendered into HTML and displayed. Currently there is no ability to customise this page which may not be to everybody's taste. The advantage is that the page can be rendered according to the application that access it. e.g. a mobile phone. 
  • We are currently in the ‘Landrush’ registration period which costs serious money. (~300USD for 3 years), at the end of march there will be a general availability which will cost the usual ~20USD
  • There really is no mechanism for selling the domains, so it could turn out to be a trademark nightmare.

Summary and thoughts

So I guess the long and short is that it looks like there is potential, and if you want a chance of ever getting the domain name you are after, register NOW. (In the case it turns out to be a waste of time you loose your years registration ~20USD) It is a much better opportunity than trying to get your domain for cyber-squatters!

I shall be ordering my .tel as soon at the landrush is available.

If you combine the concept of permanent internet connections (i.e. Mobile Broadband) with this .tel then it is easy to imaging your phone book list listing  the .tel information, when you ‘dial’ the device decides if VOIP or GSM are most appropriate. Taking it a step further then sending a message will decide if SMS, email are preferable based on the recipients’ profile, or even location.

As a complete stab in the dark I wonder if you would use it to provide people with access to your public key uses to encrypt communications?

As people want to extend the data in the .tel and include more features that it was not intended to be used for, it is easy to imagine that the end product will be Facebook without the pictures. I guess the question has to be asked, Why not just use an existing social networking site?. The creators may argue that this will be global, does not require a website or web knowledge. .tel Stores the data in an easy to access lightweight location. But in the end the problem with standards is that there are just so many of them!

Handy links

The full manual http://www.telnic.org/downloads/telguide.pdf

Some TV and radio articles (on the Telnic site)

There are *loads* of news articles.

Back to my homepage….



04:01 GMT  |  Read comments(3)

03 March

Data organisation

I have been having a look around for tools that can help organise data (thoughts rather than photos) and came across Evernote.

Whilst I am sure there are others out there I was interested to see the you can sync between many different types of devices as well as search for text in an image file.

Finding text within images is a very powerful feature. So in theory you could really go paperless, take a photo of those bills/bank statements and shred them. If this works as well as they claim then you would be able to search across your new paperless ‘office’.

Notes live in a notebook and you can tag each note with customisable tags, which are searchable.

Often when working on a ‘project’ i may see something interesting (e.g. webpage) which you can copy as an image, and add as a note. This provides the ability to build up a repository of sources for the current ‘project’

Finally there is a checkbox/to-do list capability, so each ‘project’ can have a list of tasks you wish to complete. It looks like you have the freedom to put the checkbox anywhere in your notes.

Limitations

The free version has some data limits.

Platforms

Mac/Windows/Mobile

Data formats

Text/picture (now if they could add audio ;-)

Site

http://evernote.com/

13:51 GMT  |  Read comments(1)

24 February

Cloud Computing

This is a very well written paper on cloud computing. I am not sure that I agree with the distinction between ‘Public’ and ‘Private’ clouds. In the future there will probably be multiple cloud vendors supplying clouds to a specific groups of people i.e. Defence. These are ‘Private’ clouds by the definition in this paper, but in reality they are public clouds.

 

PaperAbove the Clouds: A Berkeley View of Cloud Computing

07:44 GMT  |  Read comments(0)