Perwood Computing

Explaining it to the kids…

james — Wed, 08 Feb 2012 10:27:41 +0000

I have a hard time explaining to my children what it is I actually do, unlike some people. I have friends who are Aircraft engineers [I fix planes], Pilots [I fly planes], Shop owners [I sell stuff], Truck drivers [I drive], etc, etc. I on the other hand, write code, help build websites, run security audits, architect systems, run cost effective development teams and create development roadmaps.

It’s not easy to get it into a short sentence that a child or your family would have the attention span to actually listen to. The closest I got to a single sentence was when asked by my wife just after we got together about 9 years ago. The answer was:

“I build websites, so a sales person can sell air.”

Not that catchy I admit, but quite close to the truth back then. The reason I feel the need to try and work out how to get what I do into a sentence is that most of the time i’m just IT support for friends and family. They really don’t care that i’ve worked for Yahoo!, worked with banks and credit card companies on the first real-time authorisations of purchases on the web or that i’ve had meetings and discussions [sometimes in pubs] with founders of internet giants and inventors of languages. Can you work out how to transfer my pictures onto my new laptop? cool!

How to fix a computer

I guess it is like being a mechanic, or electrical engineer as they are now. Your skill set is such that you can diagnose a problem with engine tuning and work out that it’s the relay’s fault. But your friends ask you to come over and change the wheel, fix it up for a MOT etc.

Graph created by GraphJam.com

This is why I don’t really bother explaining what I do, I explain what the company does and how I help them achieve that.

It doesn’t help my kids understand why i’m out of the house for 13-14 hours a day, but they think it’s cool that i know the people who do the special effects for Doctor Who – so that’s a result!

And as I learnt in the past – as long as you say it with confidence and is close enough to the truth, people will accept it.

Cloud and the Edge

james — Tue, 24 Jan 2012 12:01:22 +0000

It’s not U2 splitting up and the guitarist starting a new band, but both “the edge” and “cloud” seem to be the up and coming buzz words at the moment – even though both have been around for a while, it could be that both are becoming more accepted by more traditional organisations.

CDN’s (Content Delivery Networks) have been around pretty much since the dawn of the dot-com boom, I’ve worked with Akamai since 2000 while at Yahoo! The Edge services that Akamai, Amazon Cloudfront and companies like them can really benefit companies that require global reach without the huge cost of large data centre infrastructure.

Most companies that are serving web pages use CDN’s to reduce the load on their web or database servers to allow them to have infrastructure that copes with the average load and benefit from the CDN for peak traffic times. Where companies are distributing huge files (1GB-5GB video masters) where using youtube is not really an option this is where the CDN comes into its own.

When a user from afar requests a file, the request transparently goes through the CDN, if the request has not been seen before then it starts buffering the file over to the user and transfers the file to the mirror/caching servers on the Edge closest to the user. As origin to mirror transfer is faster than origin to user, by the time the user is watching the file it is cached on the mirrors and being served from there, not the origin servers. Now with the peer-to-peer CDN model (as used by various companies including the BBC for its Iplayer) the ability to scale when a video goes viral is even greater.

So any self respecting architect, Dev manager or technical director should look into using these services where they make sense – just remember that image, JS and css file caching is a bitch when you are trying to debug issues and you forget that pesky CDN.

The Cloud is normally an easier sell:

•Reduced Cost
Cloud technology is paid incrementally

•Highly Automated
No longer do Infrastructure personnel need to worry about keeping software up to date.

•Flexibility
Cloud computing offers much more flexibility to scale both horizontally and vertically.

But you cannot just throw everything on to it without thinking. Security is an issue and from a data integrity point of view it’s true that some providers do not have good SLA’s. BUT – if you don’t have a data storage back-up and disaster recovery plan on your physical servers you may as well just throw it up on the cloud anyway, you’d be no worse off!

Security – to be honest, the main service providers (Rackspace, Amazon, Google) have better security measures in place for their data centres than you do. And if you get hacked via a web-service that could happen if the web-service was on your physical servers.

The biggest benefit for the cloud now is the automation of instances being brought online when certain criteria are hit. This takes the main pain out of constantly monitoring your systems to anticipate when you need a new slice started.

Cloud – yes you should, but please make sure you plan ahead otherwise your experience will be painful.

The Edge (CDN’s) – yes you should if you want to alleviate some load on your DB servers and web-servers to help cope with peak traffic times and overall help reduce the need for expensive hardware until you’ve adopted the cloud!

New year resolution

james — Fri, 06 Jan 2012 10:08:49 +0000

After 5 consecutive years, my New Years Resolution is not going to be giving up smoking. Not because it’s now healthy or that I feel I don’t need to give up, but rather that new year is not the time to try – too much pressure, too much expectation. That being said – there are plenty of other things to be getting on with.

I’ve been with Beam.tv [part of the Mill] since the beginning of December and I have to say that the people I now work with are dedicated and know what they are doing.

In my last post I said that you needed to have Proper Planning and Preparation to prevent piss poor performance – well part of that is being able to correctly capacity plan and have statistics and numbers in general to back up your assumptions and give correct ROI for future projects. Metrics of how users interact and use the site or service is key. Having them easily assessable, more so. Requiring developers, support teams or sys admins to pull stats from various servers is not a correct use of time. a) it should have been automated already – I cannot have been the first person to ask for such stats and b) having no central dashboard for management, sales or even key internal stakeholders means you are going to have a number of uncomfortable meetings before the truth comes out.

In a previous life – I was in a meeting where a Marketing Monkey stated to an assembled group of non web stakeholders that the new website should be redesigned as her department had done a study that said the search box should be in the margin, not where our UED and designers had put it (in the middle under the ad). I do agree – that having ads at the top of the page is open to opinion – the fact that Google, Yahoo, Amazon, Facebook all have the search box in the middle at the top means that her comment was just plain wrong and stupid. I was able to come up with any number of reasons as to why it should not be done, but the claim of metrics trumpted common sense at that point.

After the meeting, when asked for, the metrics never came – as she made it up. We were able to dodge that bullet. But had we been prepared with our own statistics i.e. number of searches, top search terms, number of NULL searches etc… we could have closed down the argument then and there in public, as opposed to later on in private.

SO – what does all of this have to do with new years and new jobs? well I want some metrics and numbers that I can call on easily and not be reliant on some chap I am now taking away from doing something important.

You think the website is bad quality?? prove it – where are the stats… You think that feature X should be removed because it’s not used?? why? what reasons/stats do you have to back up your claim?

6P and KISS

james — Fri, 18 Nov 2011 09:59:03 +0000

When you have an idea, the next step is planning. This part of the process is harder than you think, any number of projects have failed because of it (or lack of it). There is also the desire to complicate matters, it’s not a fault as such, but more a fact of life.

6P comes from a British Army adage: Proper Planning and Preparation Prevents Piss Poor Performance – as you’ll notice that is 7P’s but apart from making sure there is coffee and coke (or diet coke in the case of some engineers) preparation is pretty simple!

Proper planning comes in many forms from making sure that the SEO of the site is done correctly from URL creation to making sure that the descriptions are all unique to the IA of the site. Information Architecture of a new site is very important and takes some skill to do it right. Not only will you start inadvertently be creating orphan pages down the road but when it come to the analytics - analysing where people are going on your site – it will be a nightmare. There was a site that I worked on some time ago where the original category map was not as robust as the editors would have liked, therefore instead of grouping content in categories they started making tag pages for the content.

Just so we are clear on what we are talking about – categories are used for the vertical cutting of content (i.e. fashion > dresses, or Entertainment > Movies) in this example fashion content is split into separate sub categories to allow the user to drill down. Tags on the other hand are used for horizontal cutting of content (i.e. Madonna) a tag page would collect content from all categories both movies and dresses and allow the user to see all Madonna stories regardless of the categorization.

So what happened to the site that started using tags instead of categories? well when it came to analysing where the users went and how successful the content was – we couldn’t. Using Omniture and Google Analytics we saw bad [low] stats for all categories and couldn’t tell how the users were using the site – a fundamental problem!

Now this highlights a couple of issues.

1) Editors do not care about how their content arrives on the site as long as it appears where they want it and it only becomes a problem when you try to analyse how your site is doing.

2) The Development team should have made it impossible for people to create content without a category.

What is the learning? Development teams should not assume that people will use the product they create as they expected and if the IA is not done in consultation with the editorial department then they may just work around it. So Proper Planning Prevent’s Piss Poor Performance.

KISS – Keep it Simple Stupid. This should be used always! in the case above – a complicated category structure will only confuse users (both internal editors and external users)A six level hierarchy may mean you can see in the analytics the granular nature of how people are using your content, but it will lead to areas of the site that are very empty and it will take users ages to find content (as well as spiders from the search engines). This is BAD as users don’t like clicking too much, 3 levels is about right, maybe 4 at a push. When working in eCommerce with some very big brands, I always told them – a user needs to be able to purchase within 3 clicks from the page they entered. This principle should be used for content as well.

Why are search engines so popular? Because people do not want to navigate around the site to find content. It is as simple as that!

So – you need to plan – make sure that the IA and SEO are done correctly, in a timely manner and you have consulted the correct teams and to Keep it Simple, so you (and your users) can then find the content you have created.

Starting a start-up

james — Fri, 11 Nov 2011 12:35:43 +0000

There is a process when you are creating a new business or website for the first time that you need to go through to get through to launch. But they all start with an IDEA. It does not have to be an original idea [it cannot however be a rip-off!]or one that you made up on your own – friends and partnerships work as well as one man bands – it just has to be an idea.

The reason I say that it does not have to be an original idea is that you get a lot further by standing on the shoulders of giants and you just have to look at three of the most successful companies in the last 10 years who have re-invented what was already there. Google did not invent “search” they had an idea to make it better. Apple did not invent the mp3 player, phone or tablet, they just had an idea on how to improve it. Facebook did not create the social media platform, they created a better user experience and the rest is history. If you think you know how to improve something, then get out and do it!

I’ve been looking and advising on a number of start-ups recently and the common denominator for them are that I really did wish I came up with that idea, as it is obvious when you think about it and that the idea itself is quite simple. Some of these sites will require the luck of being “first” or getting the most traction when it comes to social media or you could call that luck. But the others just require some good planning and time and in theory they should become a success.

next time: KISS and PPP (Keep it Simple Stupid and Proper Planning Prevents Piss Poor Performance)