@jornane_no @Warcop I think we're mixing up two different cloud use cases here, which have similarities but also huge differences. We also have to look at the size of the company we're talking about. 1/
@jornane_no @Warcop A: You need infrastructure to run your product/service. Yes, either way, you will need people who develop the software, deploy it somewhere and operate it there. IMHO companies should focus on their distinctive features and buy the commodities from suppliers. 2/
@jornane_no @Warcop The smaller your company is, the more this matters. Like a small business would not invest in their own electricity infrastructure and a whole electricity department because it's obvious that it is cheaper to rent an office with batteries included. 3/
@jornane_no @Warcop If your company has special needs (e. g. you want to power your office from your own solar roof) you might invest into own infrastructure. And if you're a power plant, there's no way around own infrastructure because you're part of the public infrastructure. 4/
@jornane_no @Warcop But as long as you're making your money with what runs ON electricity (as opposed to "runs THE electricity"), someone else will do it better and cheaper because electricity is a #commodity and a lot of vendors are specialized to do it good and economically priced. 5/
@jornane_no @Warcop I recommend to read @swardley's work and try #WardleyMaps to figure out if IT infrastructure is your distinguishing feature or just a #commodity that you should buy/rent from someone. If you buy it, look for the highest abstraction possible! 6/
@jornane_no @Warcop @swardley My favorite cloud service is the satellite up-link you can rent on #AWS. aws.amazon.com/de/ground-stat… If you put a satellite into space, your main concern isn't the communication. It is just a necessity to have for what you're actually trying to do (e.g. collecting weather data). 7/
@jornane_no @Warcop @swardley Another great example (although not a cloud service) is #SpaceX Rideshare. spacex.com/rideshare/ Unless you're #NASA who provides such services, it doesn't make sense to build all that complicated stuff for yourself. And even NASA can't compete with SpaceX anymore. 8/
@jornane_no @Warcop @swardley B: You need tools that enable your business. Office, communication, collaboration, etc. If you're in #BigCorp, the main difference to cloud infrastructure is that you're most probably victim of #Microsoft's monopoly. So it will be expensive either way 🤷 9/
@jornane_no @Warcop @swardley For small business it still doesn't make sense to hire your own people to run your own instances, renting it will always be cheaper. And in my experience with #BigCorp, the main difference between #OnPrem and #Cloud is the quality of service. 10/
@jornane_no @Warcop @swardley With cloud you always have the latest features and high availability from all over the world. Two things that are. You can also scale it up and down as needed. 11/
@jornane_no @Warcop @swardley If you're starting to build up your own tool infrastructure on-prem today, you might be able to provide it at the same level as cloud. But look at the whole lifetime of the tools, you might end up where a lot of big companies are today: 12/
@jornane_no @Warcop @swardley Sooner or later a controller will figure out that this is not what the company is making money with, so they'll cut cost. Also the bigger the company is, the more complicated it gets to upgrade to a new version. Integrations with other tools, trainings, new bare metal, ... 13/
@jornane_no @Warcop @swardley A lot of complex and expensive stuff that gives you reasons to delay upgrades or skip versions. But deteriorating tools will gradually lower your company's performance (and you won't even notice it) until your employees rebel against your IT or just silently leave. 14/
@jornane_no @Warcop @swardley Example: At my former company there were ~10 tools that I used every day and ~10 more that I used occasionally. Some of those tools 5-10 years old, which caused a lack of productivity. 15/
@jornane_no @Warcop @swardley Most of those tools weren't available for at least 1 full work day/month because of bad #QoS (~0.97 availability). Internal #SLA often required only 0.95 availability (which is ridiculous!) Outages usually lasted half a day or longer. 16/
@jornane_no @Warcop @swardley 0.97^10=0.75. So there were 1-2 days per week that I had to do workarounds (e.g. noting stuff in Notepad because Jira wasn't available, later remember to put it into Jira) instead of focusing on the work I had to do. Huge productivity loss! 17/
@jornane_no @Warcop @swardley After #O365 was introduced, I remember one day that we had an outage of this size. It was the day the #Covid19 lock down started and the usage on O365 spiked. #MS had mitigated within hours and scaled up to fully fix it within 2 days. Could your #IT department do this? 18/
@jornane_no @Warcop @swardley This is an extreme example, but exactly what happened at my former company (also a reason why I left) and it is something I see happening a lot at #BigCorp right now. 19/
@jornane_no @Warcop @swardley It might be a specific symptom of the #SoftwareTransition that is currently going on in classical engineering companies (which still do not understand the value of IT) and might be less of an issue in real software companies. It might also change in the future. 20/
@jornane_no @Warcop @swardley But unless you're perfectly certain that your #IT will be provided with the necessary budget and resources to keep the quality of your infrastructure up for the next 10+ years, I would not start to host tools like office etc. on-prem today. 21/
@jornane_no @Warcop @swardley Another thing I haven't talked about is expertise (for both use cases). Does your company actually invest in building up the same expertise than a specialized provider does? If your #IT thinks they can do it better than them, it's often hybris. Eg. in #Security. 22/
@jornane_no @Warcop @swardley #OnPrem isn't more secure if you need heaps of additional infrastructure (own VPN, own SSO, own Proxy, own 2FA, ...) to operate it, which can each be a single point of failure wrt. availability as well as security. 23/
@jornane_no @Warcop @swardley This infrastructure is often considered as "just there" (like the IT people who run it). But it's an investment you have to make and keep making that you don't have if you fully switch to cloud providers 24/
@jornane_no @Warcop @swardley Where - of course - you have to pay for it, but is it really more expensive if you consider the real cost? It's usually the transition that is expensive, when you have to support both worlds in parallel, not the actual operation. 25/
@jornane_no @Warcop @swardley Bottom line: it depends on your company if #Cloud is better than #OnPrem. But most companies that think they can do it better/cheaper than specialized cloud providers should think twice if A) they focus on creating product value and B) consider the real cost of on-prem 26/26
@jornane_no @Warcop @swardley Correction: two (or even three) things that are crucial for your company's productivity
@jornane_no @Warcop @swardley Please continue here:
https://twitter.com/frederikheld/status/1542103568303493123😉
• • •
Missing some Tweet in this thread? You can try to
force a refresh
