Last tuesday I attended the Cloud Communities Day event in Milan, hosted by EuroCloud Italia. Its goal,fully successful, was to put let the public discover and put into contact with each other the various User Communities working on Cloud Computing: in addition to some of us representing the Italian VMware User Group, there were also people from OpenStack, AWS and Microsoft Azure.
It ‘was a great opportunity for discussion, and as always the best time, in my opinion, has been the short discussion at the end of the event. I had the opportunity to listen to different points of view, from users, consultants, providers, using different solutions; very interesting, although I had to note that unfortunately there are still some distorted visions of the phenomenon of Cloud Computing, and some positions some users and (especially) professionals that I fear sometimes can alienate users. This post is not meant to be a sterile criticism of these ideas, partly because it is not said that I am right, but rather an additional opportunity to present my ideas.
1. The problem is Legacy software
The statement itself is totally true. Monolithic softwares born to be executed on centralized infrastructure simply do not work with Cloud Computing, totally based on concepts like Scale-Out, Autoscaling and the like.
From this point however, the idea has been quickly distorted in ideas like:
” the fault is in the customer IT Manager not assuming his responsibilities to try and migrate their software from mainframe to distributed environments ”
” you have to invest in rewriting all that obsolete code ”
and many others with the same mood. The problem is, the IT Manager is not a coward or someone only trying to “guard his own garden” by avoiding any innovation, rather consultants often overlook the huge costs and unknown results of these kind of activities, even if they are worthy. I know for my work several banks (we all end up talking about banks and their mainframes…) that seriously thought about moving all the code inside their mainframes towards distributed systems, and almost all of them gave up because at the end it was cheaper to go on paying IBM support, and the outcome was too uncertain to put this highly critical infrastructure at risk.
Also, I often hear these statements coming from consultants, and few times from end users. In my opinion the error lies in trying to offer Cloud Computing at any cost, and pushing users to adopt it even where is not suitable. Even if everyone says the opposite, the primary reason (not the only one) to adopt Cloud Computing is savings; until you can prove to the customer that moving to the Cloud is convenient, the customer will be wary of doing so. And h will always have a ton of good reasons. In addition, there will always be systems which by their nature will never be moved to the cloud, and is detrimental to try even here to sell “Cloud” to everyone.
2. Cloud is good for everyone
Again, the statement is true, but the way it is declined often contains fundamental errors, in my opinion dictated by the same reasons I explained above. Cloud is not suitable for everyone, at least if we are talking about it as outsourcing IT services to a third-party company. One sentence in particular made me cringe:
“Some companies are spending millions for their IT budgets. Think about how much they would save by using Cloud”
There is nothing more wrong in this sentence, and I’ll show why. First, if Company X spends many millions for its IT, there are maybe areas where it could optimize expenses, but certainly a large part of the budget is proportional to the size of the infrastructure they have to manage. A mass move to the Cloud can not in any case reduce that budget to one-tenth, but it will maybe bring some degree of optimization.
Second, the migration to Cloud has a price. And if the infrastructure is that big, probably also the migration costs will be notable. Are we so sure, from a financial standpoint, there will be sooner or later a breakeven between migration costs and savings? Maybe.
Third, and most important for me, some infrastructure are so big, and have so many skilled guys managing it, that the most obvious choice for those companies will be for sure a Cloud, but a private one! I thought about this while listening to the OpenStack guys, while they were talking about the “end user” using their platform: NASA, CERN, MIT… Awesome use cases to impress the attendees, but totally useless from a business point of view: those are institutions with such a size and internal skills, that they basically made their own Cloud, and they probably contributed to some of the code of OpenStack to fulfill some of their needs.
Can you use those examples to persuade a manufacturing company to move the a (Public) Cloud? In my opinion NO.
3. Security is not only Availability
Probably is because of my past in IT Security, but every time I hear someone talking about Security I feel the message is always incomplete. there was a simple question during the panel about the security offered by those public infrastructures, and many replies with the usual example:
“How can you compare the reliability of a datacenter where every component is redundant, with highly skilled staff managing it 24 hours a day, with your single rack with a small UPS and an ADSL line, where noone is guarding it after 6 pm?”
Put it this way, the answer is foregone, and even better makes the asker feel a little bit dumb. Coming from IT Security, I instead need to say to those guys giving these kind of answers, you are wrong, because Security is not only Availability, but also Confidentiality and Integrity:
In particular, the doubts coming from users are towards Confidentiality over their data, especially when those data are under the control of another entity. Trying to dispel the doubts about Confidentiality by talking about Availability is a stupid move. Instead, we should talk about data governance, but if you have too much “hurry” to sell Cloud, you obviously tend (maybe unconsciously) to go beyond these doubts. But until they are not dispelled, customers will have every reason to be wary.
My message is simple: Cloud Computing has many undoubted advantages, they cannot be applied to every situation, and “it works” only if you explain clearyl to users all the pros and cons, so they could make a informed choice. Talking about it in an approximate way for the sole purpose of “selling” is likely to turn people away.
I do not want this post to appear cynical or too critical, but simply during the event I had little time to expose these ideas (I recommend for the next event, please moderate interventions, otherwise it becomes an arena). If you want to reply, feel free to post a comment.