Technology
Cloud Computing
Software Development
Kubernetes
Private Cloud
Legacy Systems
Cloud Transition
Cloud Evolution
Cloud Revolution
IT Infrastructure
X : You're not a fan of private cloud or kubernetes?
Me : They have their uses.
X : Really?
Me : Yes. Private cloud and kubernetes matter if you’ve got legacy systems and you need a place to strangle them. But it’s not the future and you don’t build new there.
Me : They have their uses.
X : Really?
Me : Yes. Private cloud and kubernetes matter if you’ve got legacy systems and you need a place to strangle them. But it’s not the future and you don’t build new there.
X : What about as a transition play to cloud?
Me : Be careful there. The longer the bridge you build between where you are and where you want to be then the less likely it becomes that you'll ever finish crossing it.
Me : Be careful there. The longer the bridge you build between where you are and where you want to be then the less likely it becomes that you'll ever finish crossing it.
X : Cloud is evolving.
Me : Everything evolves. What you're hoping is that it magically evolves so that your legacy just moves there. Alas no. You'll have to re-architect. So yes, cloud is evolving but you'll need a revolution to move there.
Me : Everything evolves. What you're hoping is that it magically evolves so that your legacy just moves there. Alas no. You'll have to re-architect. So yes, cloud is evolving but you'll need a revolution to move there.
X : How long does it take to move to cloud?
Me : Depends upon context i.e. the state of your business, what business you are in, the size and the zeal of the execs. It took Netflix 7 years to go data centre zero, Capital One took about 8 years ... 7-8 is a reasonable benchmark.
Me : Depends upon context i.e. the state of your business, what business you are in, the size and the zeal of the execs. It took Netflix 7 years to go data centre zero, Capital One took about 8 years ... 7-8 is a reasonable benchmark.
X : Zeal?
Me : The willingness or not for half measures. Be careful of transitional steps they have a tendency of becoming permanent especially in organisations where inertia exists which is pretty much everywhere.
Me : The willingness or not for half measures. Be careful of transitional steps they have a tendency of becoming permanent especially in organisations where inertia exists which is pretty much everywhere.
X : But it's a lot of work.
Me : It can be but going from a traditional business to one that can design, build and deploy a new call centre system in 72 hours is never going to be a "click button to accept" exercise - capitalone.com
Me : It can be but going from a traditional business to one that can design, build and deploy a new call centre system in 72 hours is never going to be a "click button to accept" exercise - capitalone.com
Me : Look, I can understand the concerns if you're a startup that for some reason decided to build a data centre five years ago. But if your company is a decade old, has lots of legacy then you've already had plenty of time to migrate - same as Capital One and Netflix ...
Me : ... you might not have migrated but that was your company's choice. Hoping to find some transitional step to make the process is easier is ... well, wishful thinking.
X : Any advice on building a private cloud?
Me : Don't. What are you after?
X : Useful metrics?
Me : If I look at a private cloud offering, I'll go straight to estimated operating cost per VM and ask "what % of this is electricity" ... the answer should be about 40%.
X : Eh?
Me : Don't. What are you after?
X : Useful metrics?
Me : If I look at a private cloud offering, I'll go straight to estimated operating cost per VM and ask "what % of this is electricity" ... the answer should be about 40%.
X : Eh?
Me : That's my first question. The second would be, "What % of this is the depreciated cost of hardware over three years" to which the answer should be around 10-15%.
X : And if it's not?
Me : Then I start getting very suspicious that not all the costs are included.
X : And if it's not?
Me : Then I start getting very suspicious that not all the costs are included.
X : I'm not sure I understand.
Me : If the server costs $4,500 and depreciates over 3 years then the running cost of the server should be around $10,000 per year ... as a rule of thumb. Hardware is the small part.
Me : If the server costs $4,500 and depreciates over 3 years then the running cost of the server should be around $10,000 per year ... as a rule of thumb. Hardware is the small part.
X : That's multiples of the hardware.
Me : Oh, yes. You learn that running data centres - from opportunity cost of capital to power to aircon to replacement costs to failure to physical security to buildings to racks to internal networks to UPS to ... it goes on and on.
Me : Oh, yes. You learn that running data centres - from opportunity cost of capital to power to aircon to replacement costs to failure to physical security to buildings to racks to internal networks to UPS to ... it goes on and on.
X : Staff?
Me : Yes
X : Utilisation
Me : Yes. That last one ratchets up costs when talking about VMs. Anyway, check these costs because people often forget to include them in the business case.
X : How often?
Me : I haven't seen one in fifteen years which had all the costs.
Me : Yes
X : Utilisation
Me : Yes. That last one ratchets up costs when talking about VMs. Anyway, check these costs because people often forget to include them in the business case.
X : How often?
Me : I haven't seen one in fifteen years which had all the costs.
X : Public cloud?
Me : It's all included. No-one sends you an electricity bill for your use of public cloud or a bill for replacing a stack of UPS batteries that have died.
Me : It's all included. No-one sends you an electricity bill for your use of public cloud or a bill for replacing a stack of UPS batteries that have died.
X : And when 50% of your cost is meaningless to your end goals.... What value are you really looking for?
Me : That's a question on where you want the energy and effort of your organisation focused ... I tend to view it's best to have it focused on the user need.
Me : That's a question on where you want the energy and effort of your organisation focused ... I tend to view it's best to have it focused on the user need.
Me : i.e. If I'm an insurance company then I'm going to focus my effort on insurance and not building infrastructure clusters. That would be like a hospital that can't treat patients because they're too busy home brewing penicillin.
Me : Anyway, we learnt these lessons about 10 -12 years ago, circa 2009 - 2011. It seems really odd to be having them today. It is a nice trip down memory lane but then it's 2021 after all.
X : So, there is no point in private cloud?
Me : No. It can be useful if you want to strangle some legacy. However, you need to start asking why you're still in a position that you would contemplate it. You've have had over a decade now, that's a long time to dally about.
Me : No. It can be useful if you want to strangle some legacy. However, you need to start asking why you're still in a position that you would contemplate it. You've have had over a decade now, that's a long time to dally about.
X : Strangle?
Me : Strangler pattern.
Me : Strangler pattern.
X : Cloud is over a decade old.
Me : Of course but you need to give people five or so years to get used to the idea, see others move there i.e. 2006 - 2011 is when people should have been investigating, 2011+ is when you should have been motoring. Some like Netflix started early.
Me : Of course but you need to give people five or so years to get used to the idea, see others move there i.e. 2006 - 2011 is when people should have been investigating, 2011+ is when you should have been motoring. Some like Netflix started early.
X : Now is a good time to move to cloud?
Me : No, it could be a dreadful time. Let us say you decide to head out on an IaaS + DevOps journey today, it'll take you 7-8 years at which you point you've built the new legacy to migrate from. Better to start today with Serverless.
Me : No, it could be a dreadful time. Let us say you decide to head out on an IaaS + DevOps journey today, it'll take you 7-8 years at which you point you've built the new legacy to migrate from. Better to start today with Serverless.
Me : When you think about migrations, look at where you and where the market is going to be at the end of your journey. You've had plenty of time to investigate serverless i.e. 5 years+ (2014-2021) ... now you should be motoring.
X : 8 year plan to move to serverless?
Me : Bingo.
X : 8 year plan to move to serverless?
Me : Bingo.
X : So become the Capital One of serverless?
Me : Exactly. Except Capital One isn't daft and so I suspect they'll be doing that as well. But you're at least playing in the right half of the pitch.
Me : Exactly. Except Capital One isn't daft and so I suspect they'll be doing that as well. But you're at least playing in the right half of the pitch.
Loading suggestions...