Blog RSSReliance on 3rd Party Services – Good or Bad?
January 20, 2012More and more, developers are becoming reliant on 3rd party services to provide data, provide code or enhance the site that they offer. This can be as simple as somewhere else hosting some of your resources, be it simple Javascript libraries, or actual file hosting (via a Content Delivery Network, or CDN).
Now, I can see the benefits on resources to offload heavy items to other services and while those services are up and operational, its brilliant. Saves you money and means you can serve more users from the same hardware.
However, what happens when they do go down? In most cases, your site should still work just missing functionality. Now, in most cases you will be pre-notified if there is likely to be any prolonged downtime to allow you to do something about it.
The thing that prompted this article was the shutdown of MegaUpload by the US Federal prosecutors. They have had MegaUpload shutdown instantly for violating piracy laws, however, sites like this are used for genuine reasons too – the users have had no time to download any files they may have uploaded there and host them anywhere else. The service was just taken straight offline. This has meant that many sites with hundreds of thousands of legitimate files hosted on this service now point to dead links with potentially no way to get the files back.
What would happy to you if Dropbox suddenly went offline? Or another cloud based file storage solution that is out of your control?
New Servers – Costing them Out.
January 9, 2012When planning server upgrades and features, cost is always high on my personal consideration list. This isn’t just an upfront cost or a monthly cost but what gets called the total cost of ownership. What you pay over the life of the box.
If you look at buying servers the prospect of up to a few thousand pounds upfront can seem a lot. However think about how long the hardware may run for. It may not be as bad then, rented servers are nice when you need something quick or plan to upgrade often but as soon as you start to look at keeping them for longer periods of time then you can pay way over the odds for them.
In this case, recently I have just bought myself a cheap HP server to host my personal websites on (this blog, www.ontvnow.co.uk and a couple of other sites), for £500 I can get a Quad Xeon server, 8GB RAM and Dual 1TB (RAID 1) hard drives. This gives a nice light weight server to handle simple web serving and doesn’t cost the world.
I will probably keep this machine running for 3 years, so add in colocation costs for a single U of about £70 / month for 36 months that equates to about £3000 over the ‘life’.
As soon as you look at servers elsewhere of a similar spec, you are talking at rents of £250+ a month for a similar spec machine, giving over the same 3 years a cost of £9000. Potentially three times the cost for the same hardware, but at the end of that, you don’t own anything.
If you have the ability to run it, suddenly it becomes quite cost effective to colocate a server. Maybe more so than quite a few realise.
Even Microsoft Wants to rid the world of IE6 now
March 5, 201110 years ago now, Microsoft unleashed Internet Explorer 6 on the world… even now it is still a heavily used browser, even after 2 upgrades have been released and a 3rd is on the way this year.
Microsoft themselves have now created a site to countdown to the final death of the browser http://www.ie6countdown.com/ This site is intended to encourage users to finally upgrade from Internet Explorer 6 to a more recent browser.
There are major benefits to doing so, like a much better and more secure internet experience however a lot of large corporations are still stuck on Windows 2000 which can’t have any newer version than IE6.
Is it good to work without a framework now?
March 3, 2011My initial thought would be “god no”, however there are some benefits to working without one.
The biggest advantage to what I’ll call stray code, without any form of framework is its easy to add to, anyone can jump in and work on it without needing too much experience of how the code works. Pages can run separately from each other without any issues however this does itself lead to some issues. Is all the code constant? Answer… probably not.
One thing I notice, the more time I spend working with frameworks, is framework lock-in. What I mean by this, is once you’ve written the code for one specific framework, it isn’t re-usable in another due to specific modules and objects that have been used within that code.
Having a framework and suitable documentation does make it easy for any competent programmer to jump in and work, more so if a large known framework is used, for example Zend Framework, Symphony etc. These also can be requirements for knowledge when employing new staff.
As someone who spends all day using a custom made framework, when the requirement comes to write none framework code, I really have to think about it. Database calls, yep… I’ve got an object for that, I now have to think about mysql_* calls for example.
What becomes a better solution, rather than a framework, is a collection of libraries. These are code blocks / objects that can easily be swapped out without interfering with the running of the code.
While I can see the problems frameworks generate, it doesn’t at all mean I’m going to drop using them – they do save a lot of time for development, however I can see issues that do arise from their use.
Facebook Visualising Friendships
December 14, 2010You can do interesting things with data, in this case a Facebook engineer has taken all the Facebook friend relationships and mapped them out producing the following image.
Click here to view a high rest image of this
Every relationship was rendered as a line between the 2 points that those users were registered as being located at. The above image is only this data but you can quite clearly see an outline of the world.
To find about the process used to create this graph, http://www.facebook.com/notes/facebook-engineering/visualizing-friendships/469716398919