Cloud Expo #4: Agile Development, CMOs and the Borg

The fourth and final day of the conference. I sat through a few higher-level, vendor pitches today which were ok but they didn't introduce anything blogworthy.

The “interesting” section today is the section Cloud Scar Tissue below.

Agile Development in the Cloud - Jan Aleman, Servoy

I wouldn't normally attend a session with 'Agile Development' in the title but this level of the stack seems to be very neglected. As I've mentioned in previous posts the focus seems to be on the infrastructure and platform layers. According to Jan:

• 65% of new offerings are SaaS
• 30% of total market is SaaS
• 70% of all software is on premise.

So, why not write your apps to run in both? Again the speaker reiterated that as a developer IaaS is of no interest to you, all you really care about is the PaaS layer. (Force.com and Azure are the leaders in the PaaS space.) PaaS gives you the multi-tenancy, security, I/O support etc. As a developer though, how mature are their development offerings? Are you locked in? Can you develop one code line and have it deployed into on and off premise.

I was hoping that this session would be a little bit more of an academic presentation about how agile development works in the cloud but it really was just a marketing pitch for their agile development platform. They use Eclipse as the UI and he did build an application in real time during the session which he published out to the web and we were able to use it from our iPads/laptops. That was neat but I didn't learn anything to take back to the office.

CMO Panel discussion

How much of a role does branding make in a Cloud company today?

One of the panelists said that none of the companies on the stage was big enough to have a real 'brand' technically. The challenge that I think they have is communicating their value in a way that people can understand. Marketing via t-shirts seems to be a valid and effective approach that they are using right now just to gain awareness. They are in the stage of just standing out and raising awareness.

The topic of Bloggers arrived. They see the blogosphere as a place to take your brand (it is agnostic, cheap and a leveler). In the blogosphere, 'cloud people' are relatively well behaved and it is fairly flat. For example, with security blogs we get huge spikes as things happen but in the cloud blogs it spikes less.

The most interesting thing to me? The marketing guys were very pragmatic which is probably because the companies are small and they are stuck the technical intricacies right now. I'm sure they'll change as their companies grow and they forget their roots!

Cloud Scar Tissue: Real world implementation lessons learned from the early adopters. Anthony Skipper, ServiceMesh

Best session of the week IMHO. Anthony walked through a comprehensive list of lessons learned in deployments. He had issues but he also had answers!

One big thing though...he said cloud provisioned systems are like the Borg, each one is the same as the others...what? Non-geek alert, intruder detected. I thought that the guy with the mullet and sandals in the second row was going into apoplexy. Dude, the Borg were not clones, they looked really different but were part of a common hive. I'll overlook this just because the rest of your session rocked but if you are going to quote Star Trek at a cloud convention then tread carefully my friend. ;-)

So, back to the session...Anthony listed the main areas where customers have issues. I'm not doing it justice but here are my highlights:

Not believing that org change is needed

• IT fiefdoms. IT and the business are not built around the same structure as the cloud. Your cloud implementation is likely to cross the existing organizational boundaries and IT needs to be prepared for this.
• Operations will be much closer to the developers and will have good feedback for engineering. Again, the organization needs to be prepared for this new relationship.
• Architects will be forced to come down from their ivory towers and become more 'real-time'. The effects of their work can be seen almost immediately and that feedback loop will change the way they think and work.

Boiling the ocean

• This was not about how quickly you can get to the cloud, it was more realizing that you are not going to move everything to the cloud, some systems may never move. Greenfield apps will have a much better chance to survive out in the ether but some of your existing apps will not. You need to decide whether to live with this, rebuild the apps or change your business processes.
• The more service providers that you use the more connection points you are creating. You might just want to have one...or as few as necessary.

Failure to simplify

• We've all heard the adage that "complexity kills". It is not uncommon for a basic VM to have between 80-150+ integration points. When you move to the cloud those integration points are still there but now they are away from your existing infrastructure.
• Think about this level of complexity and then realize that a decent sized company could be spinning up hundreds of images a day. If you have too many integration points or complex relationships then this may not be manageable.
• The integration points may not be automatable, if any part is not automatable then the entire process becomes a burden on someone in IT again.
• In the world of enterprise software this is very important, especially with legacy systems. Can you simplify the integration points? Can you fully automate the provisioning?

Vendor management relationships

• Negotiating contracts with cloud providers should not be underestimated as a task, especially for a large scale contract. This can take 6+ months just to negotiate the deal and this has been known to actually stop a project.

Believing the vendor hype

• Customers believe too much of the story. Companies claim to "cloudify" any existing app in minutes with a new widget. If this sounds like a reasonable claim then email me because I have a bridge that I'd like to sell to you.
• Watch out for cloud providers with their own management tools...sure "as eggs is eggs" those tools will not work across platforms.
• This was a very long list of things not to believe...maybe it would have been a shorter list if it was exclusive not inclusive!

Cloud addiction (consuming too much and not knowing when to say no).

• Think about how you are billed. If you cannot predict those billing factors then your costs could balloon. Is it I/O, storage, compute cycles?
• There are solutions that are cost effective in the cloud but as they grow they may be more effective to have on premise.
• People find it hard to believe that it might NOT be cheaper to deploy in the cloud because of the hype but the cloud is cheap because of scale...once you have that same scale in-house then you might not see the savings.
• Have a policy (that word again) to review whether the cloud solutions are cost effective

Failure to implement policies from the start.

• This is just a great opportunity to drive process improvements.
• don't forget that just because it is technically possible to move to the cloud doesn't mean that it is legally possible.
• What to do if a bright spark in the business starts his/her own cloud deployment armed with a credit card and a browser? What's your policy there?
• The policy list seems like it could go on for a long time so here's what I'd say. Moving to the cloud is like any other piece of technology. It will only facilitate the running of your business, it will not drive it. You need to know what you are trying to achieve.

How is it possible that Security wasn't on his list? He said that there are security solutions to most of the problems and he reiterated that moving to the cloud may improve security.

Comments

[Andrew Chapman]

The ServiceMesh team have posted Anthony Skipper's deck if you are interested in the details...

http://www.servicemesh.com/wp-content/uploads/2011/06/ServiceMesh-Technical_Session_at_Cloud_Expo_2011.pdf

Enjoy

[Florian]

Hello Andrew,

thanks for this interesting post. IMHO the "Vendor management relationships" plus contracts negotiations is the most critical part in any cloud strategy. As soon as the cloud offering is covering multiple locations a cloud strategy can get very difficult. Lately I have been on the ACC (Association of Corporate Counsels) Conference. There one hot topic has been Privacy Laws and Data Processing Laws, because in Europe these laws are much more stricter than in the US. Even certifications like SAS 70 Level II and ISO 27001 will not really help you at this point.
Personally I see a lot of tech talk around cloud computing. Usually most people forget the contracts between the service provider and cloud consumer are very difficult to manage and negotiate.

Regards,

Florian