Now I’m just using Amazon as an example here (although they do seem to lead the way in this at the moment) – advice should apply elsewhere as well.

So what’s the deal with Amazon EC2 (‘cloud server’) pricing? This is supposed to explain it in more simple terms:

1) You have regular pricing. Just launch an EC2 instance and pay-per hour. Couldn’t be simpler. When to use: For occasionals & test machines.

2) ‘Reserved instances’ – or pre-paid. Sort of. Here you pay a lump-sum up-front for the ‘right’ to purchase at cheaper hourly rates. You MUST use this above option (1) if you are planning to run your instance (or any combination of instances) for longer than 5 months (in total) per year. So if you launch up an odd instance, shut it down a week later, then only do that again 2 months later, don’t bother, ‘Reserved’ will be more expensive. But remember: the cheap price can ‘hop’ from one instance to another, meaning it doesn’t have to be the same instance running that benefits from the reduced pricing. Amazon just mixes & matches to make sure you get the cheapest rate.
(NOTE: is DOES matter which instance type you use. So a Large instance is not the same as a Small instance – buy a reserverd ‘Small’, and reduced charges will NOT apply to your large instances)
When to use: For operational systems or very long pilots/test/acceptance.

3) Spot instances. This is new. This is a bidding system that presumably makes use of CPU cycles that have been allocated by others (e.g. reserved instances) but are not being used. This means you can get potentially much lower rates.
However the instances will only start when you price is met, meaning you can’t run immediately in all cases. And Amazon will shut down your instance when the bidding price rises above your maximum. This means that your application must be able to handle being cut off at any point.
When to use: Batch processing that you have made rock-solid.

Hope it helps clear things a little. For the procurement department: Managing (2) and (3) can really bring down your costs.

Pauwl

To the rescue comes Soocial. Talk about making something that could be difficult easy.

Now I can get my syncing going fine with a single phone and my PC, but I didn’t even bother to help my dad try. I just sent him to Soocial for that. 1 video later and he’d got it syncing himself. Now that would be worth paying money for. It is of course, as so many services out there – free.

Syncronising contacts is now easy for all. And it’s helped me out enormously – I no longer feel obliged to help people get their sync working – I just send them to the Soocial crew.

A big THANKS to them for making my life a little bit more easy.

(The fact that we don’t have the business hassle of people walking around without backed-up phones also helps of course

(The parallel with IT applications might not yet be clear, but hang in there.)

Now this had been the first time it had been cleaned in quite a while, so it took us a good hour to go through all the steps of cleaning the air filter, the body, the guide and the chain, stopping to grease the guide and file the chain. We took it for a test-drive, then decided on a little more honing of the chain to get a more rough-cut (get through wood faster).

The second time – my expert assures me – will take only 15 minutes. Do it often enough and the machine will always work, will last years longer, and will always but well.

So cut back to a workplace where we’re using the same tools (business applications) for years and years. But we neglect the visuals, and only perform the worst kind of maintenance – patching. We patch and patch. Imaging patching your chainsaw everytime someone saw a vulnerability in theirs. Not only would you have something physically to heavy to lift, but slowly your ability to maintain it erodes.

So why don’t we move to a different approach to applications, in which we give them regular overhauls – and not only of the internal engines (without making those too heavy to carry), but in particular the visuals as well.

For those who are looking for yet another reason to use a CRM – email campaigns might be it. SugarCRM is our CRM of choice – but SalesForce.com will do you too (but read this first: SugarCRM v.s. SalesForce.com).

We do send out newsletters for clients using a different system, which does give a lot more information on recipients, but I really like the basic functionality already built into SugarCRM. You can do HTML emails, but we choose to do text-only ones when sending out to a population of IT managers. And for this, we get great results with using SugarCRM.

All the legal requirements like unsubscribing are included, so we can properly manage the source lists.

It’s intuitive to setup and use as well, which makes it easy for you to get up and running fast.

Thanks to all those that support SugarCRM.

What I found so amazing was the total lack of anything that looked like a cloud.

Storage vendors were pushing hierarchical storage, fibre-channel, and SATA. Nothin new there really, although I did see something nice from Compellent, who integrate SSD, Fibre channel, and SATA drives in the same volume. This means the longer your data stays inactive (data on a block level that is), the further down it shifts, so to cheaper storage. Nice.

But what about Rackspace or Amazon with their cloud storage solutions? NOWHERE to be seen.

Only once did I even hear the world cloud, and it was mentioned by the email service provider Mimecast – who admittedly have a nice hybrid on-of-site mail solution. If you’re looking for higher availability of your mail, look no further I think. At least, from the presentation it looked good, and no reason to believe they don’t have the actual service running well behind it – some good names of customers who already use the service.

Of course there was Dell, IBM, Sun and LOADS of networking companies. None of them however had cloud solutions on show – at least not very visibly.

Why I wonder. I have no real answer for that – and neither did any of the vendors I asked.

Is it that you only push the cloud on the cloud? Or is it seen as too cutting-edge to be at a trade show? Surely not. But I wonder.

Any cloud of any use that is. Of course everyone will agree that Web 2.0 startups can’t run on anything but a public cloud – purely for the cost. Smart (?) businesses will agree that for them there can only be a private cloud – the risks are otherwise too high.

Now I won’t dive down the mineshaft and rant on about where actual business risks lie, and that frequently it’s not about over-protecting IT areas. What I do want to do is say just a little about virtual private clouds.

Imagine: We all wake up tomorrow and Barnes & Noble (the bookstore for those not in the US) buy Amazon. Unlikely of course, but hang in there for a moment.
Now, has Barnes & Noble just moved a significant part of their business (the new part) onto a public cloud? Or has the public cloud (EC2/S3) now become a private cloud, and will everyone have to get off it?

No. It remains as it was, Amazon run their services on a public cloud, although a virtual private part of it. Just like every single datacentre of every single company. Your datacentre is (virtually) walled off from the rest of the internet, but it is most deinfinitely connected.

So you already have a virtual private datacentre, and if you’re virtualising within it, you have a virtual private cloud as well.

So what is stopping large businesses move to a real virtual private cloud? Here are 4 reasons:

1. Fear for the data. Fear of losing the data, having it stolen.
2. Fear for the service disruption. You don’t control the servers – you can lose the service.
3. Fear of the unknown. There are more than 50 definitions of what a cloud is, so who knows what you’re actually getting?
4. Fear of lock-in. Once you are on cloud A, you can’t easily go to cloud B should you chose.

Let’s debunk these 1 by 1.

1) Fear of losing data. Aside from the physical security, your data is as protected as it is in your datacentre. You control the firewalling, certificates, user-id’s and passwords. The reality is it’s just as safe on most clouds as it is in your own hands – unless you’ve disconnected your servers from the network. Anything you have now (DMZ’s, double-firewalls etc.) can be built on the cloud, so get over the fear.

2) Fear of service disruption. Bad sys admin is bad sys admin. Poor database management remains poor database management. So what happens in your datacentre is what happens on the cloud – with one exception – the layer below the OS. And here the cloud promises – and delivers – exceptional abilities to replicate servers, quickly commission replacements, and ingeneral make your service disruption a whole lot shorter. If loadbalancers go down and break, you have a new one installed and comissioned in under 5 minutes. Fear is the wrong word: Optimism about shorter service disruptions is what you should have.

3) Fear of the unknown. Large vendors are desperately trying to pull parts of your datacentres into theirs. They have been for years, calling it co-location (where in reality you are generally 1 firewall away from the sysadmin that is managing another companies systems – in the same building – from the same machine). Now they are also calling it ‘cloud’ but are doing so by spreading some FUD about the risks/uncertainties of using the public cloud for private purposes. Getting over this fear is most difficult – you need good advice that you trust. Unfortunately this will not be forthcoming from most IT vendors in the short-term, who are still looking for a traditional lock-in scenario.

4) Fear of lock-in. Talking about lock – you might think cloud will lock you into a particular cloud-vendor. At the moment; it will. But it locks you in about as much as you are locked into your current datacentre supplier (presuming it’s on their premises). Moving anything to another vendor is a real pain – and cost. But with the cloud this gets easier – if not for the simple reason of no longer having to move the physical hardware. But things are getting even better – with Eucalyptus and Ubuntu teaming up, there is now a second ‘Amazon’ compliant protocol around, and it seems like there will soon be a de-facto cloud standard. This means that soon (maybe even in 2009) you will start seeing the ability to move your instances from one cloud to another (that is across vendors), whilst maintaining connectivity and services.

Whilst not comprehensive, the list has most of the fears I run into when talking to companies on a daily basis. I hope you agree with the content. If not, let me know by posting a comment.

Pauwl Lunow

Now we’ve got a team that’s using Pivotal Tracker. For them the most natural thing to do – split work into bits of 0,2,4,8 hours, and create a ‘story’ for each.
Based on these work packages (for work needed in the next 2 hours there is no ambiguity – so it’s clear what each package is), they pick them off the ‘backlog’ queue and get to work.

We work to ‘iterations’ now, with each iteration being 40 hours per person per week (or however many that person works). There are 4 in the team, 3 work 5 days, 1 works 4, so we have 1 iteration being 152 hours (40+40+40+32).

See screenshot for a little more detail.

You can split things into releases using ‘markers’ as well, and there is basic reporting on progress of items.

All in all I can fully see why the team have chosen to use it, though the old-dog in me is still wandering around sniffing to find the Gantt chart.

I have sooooo much to learn.

Anyhow, I think we’ll all move off traditional project mgt tools to pivotal tracker, at least for all programming & web projects.

If they take longer (or shorter) to complete a package it all automatically gets updated – so I’m up to date with the current speed of work.

Check it out – also for the old dogs.

Pauwl Lunow

1) Data ownership
In SugarCRM it’s so much clearer who owns the data. WE DO.
Exporting data to move to any other system from SugarCRM is simply 1 click away for each data type.
In SalesForce you need to create reports, then export them, a process that although easy enough when you do it a few times, doesn’t create the feeling that you’re in control of your own data.

SugarCRM: 9
SalesForce.com: 6

2) Open Source
But aside from this there is also the ‘open source’ part of SugarCRM. By being open-source SugarCRM is creating an opportunity for others to enhance the product, but more importantly, also for companies to commercialize the offering themselves, allowing them to provide a greater range of services based on Open Source software.

SugarCRM: 9
SalesForce.com: 5

3) Features
Well this is a real non-issue for us. Both SugarCRM and SalesForce.com are feature loaded to the point that we only really use a fraction of the functionality. So for us feature-richness of either platform was not something we needed to look at for long. The relatively simple features we use are available in both.
From the first looks SalesForce.com has a greater number of apps to add, but all plugins we’ve considered & looked for were available on both platforms.

SugarCRM: 7
SalesForce.com: 9

4) Cost
Cost is not the point for entry-level, hosted CRM. If it’s just the basics (and no plugins) you need, they cost the same hosted. If you are looking to grow in functionality, then in my experience the SugarCRM pricing model looks better.
SugarCRM: 8
SalesForce.com: 5

The final scores:

SugarCRM: 8

SalesForce.com: 6

But of course that didn’t last. With trucks, airoplanes, and ships carrying much of the load now, the rail is just another way of getting something (or someone) somewhere.

So with the increase of a variety of transport mechanisms, and an increase in the places that could be reached quickly by a combination of transport mechanisms, distribution centres started to arise. Excellent distribution (collection from multiple sources, re-combining goods and re-sending them to multiple other destinations) has become a science. Just-in-time distribution made the headlines years ago when Dell had seemed to perfect it for computers, but realistically supermarkets have done a much better job for years.

Now we’re seeing a lot of ‘hubs’ go up on the internet. At first there where the fledgling starts such as Yahoo Pipes. Then with Facebook, SalesForce.com and all sorts of plugins started appearing, the search was on for the ultimate distribution mechanism. The ‘railroad’ has become a given. The issue is now how to get your content collected, re-packaged, then delivered to a host of end-points in the most effective way.

Of course with traditional goods and distribution the costs are clearly (physically) visible. With bits of information flowing about this is not the case. In fact, making the ultimate distribution point might actually mean making it invisible for the masses.

And with invisibility comes the expectation of things being ‘free’. What kind of business model will survive that? We will know by the end of the year I think.

Basically it goes like this: Cheap HW pushes cost to elsewhere in the eco-system. Ergo cheap HW is actually bad.

Now that might be an oversimplification on my part, but was food for thought for me again about the costs of software.
We’re big on Ubuntu, Nagios, and are moving from SalesForce.com to SugarCRM. For Ubuntu and Nagios it’s clearly about the cost of software, something versus nothing.

For SalesForce.com move to SugarCRM it’s interestingly different. We couldn’t of course buy the SalesForce.com software, so rule out moving due to a cost-of-software reason.

So what about cost-of-service then? Well yes, going for a hosted SugarCRM would make a difference, but probably not worth the pain of moving from one (really good) CRM to another (really good) CRM.

So we’re moving to actually hosting it ourselves (hence the choice for SugarCRM over SalesForce.com). Of course we’ll have to provide the support & services ourselves, but we have those kind of people, so it’s another one of the tasks to be done, we increase the scope of what our people do (which I hope will keep them happier in the long run), and actually have lower costs.

Now in that whole equation I see lower cost only – no shifting of costs to elsewhere in the ecosystem.
So the ZD-Net article was about HW. Does the parallel not exist when you shift to software, or could it be that the shift of cost actually doesn’t happen?

I don’t believe the costs shift. I believe the whole dynamics change, so almost any comparison is impossible to make.

Bottom line though for us is that free (or really cheap) means a more interested workforce, and a more personal service.