A letter from the real world

Recently, I had the fortune of stepping outside of our carefully constructed tech bubble and step into real world for few days. You know a world outside of clouds, VMs, Software Defined Stuff etc.

I sat next to people outside of tech industry; I watched them do their work. I was not a mere tourist sitting in a conference room, I tried connecting with them. I went in there expecting to find technology novices that are not very smart, resigned to their fate of using computers to keep their jobs and generally hating software. I was also told to lower my expectations and these people do not know how to use computers and they are tired and old. I am talking about end users of enterprise software, not IT folks. These are also mid-sized companies, not Fortune 1000 companies.

What I found was different. I found people full of energy, curiosity and a willingness to learn. They were not afraid of computers, but were afraid of making a mistake that breaks a computer. Whenever an hourglass was present they kept saying “Its thinking, its thinking – sometimes it thinks hard” – they accepted that computers are smart and they generally work. When software is slow, they think it’s doing some complex thing.  These people are trying to do the important work that keeps America running. They work in industries that are seen by technology folks as being outdated, stale and seriously in need of disruption.


They are supported by an IT staff that is short on budget and juggling multiple balls in the air to keep lights on. They have outdated servers that are over capacity, server rooms that are overflowing and client server software that is harder to maintain. Given past experiences of changing their software or upgrading it, they are bit gun shy. They would rather apply band aid to the problem than undertake a new software implementation. The tech industry have screwed them and left deep wounds. We can blame majority of large vendors that built crappy software in the 90s and nickeled and dimed customers with professional services. I am sure there are also some success stories, but current state is that majority of these systems have become useless for future.

This real world I speak of is now at a breaking point – the band aids are not holding things together and very few large competitors are eating into their business. These guys have a risk of going out of business in next decade or two unless they catch up and improve their workflows, modernize their systems and remove inefficiencies. The IT guys are frustrated and business guys have had it with IT not delivering. The blamestorming is not happening within IT as we vendors love to talk about, it’s happening between IT and business.

A person who is not familiar with the real world and who haven’t spent time with these people may conclude that doing a full stack startup and disrupting these industries wholesale is the right approach. This may work for few, but doing full stack startups for core industries that run US is simply not going to happen in near term. Entrepreneurs shouting full stack do not have the patience to build solid foundations for replacing these businesses. They are not willing to wait 15 years to build a company; they want to flip a company in 4-5 years for few billion dollars. Full Stack craze will die down in an year or two is my guess.

What do we need to do?

We need to offer these organizations upgrade to solutions that take away responsibility of physical systems and software away from resource constrained IT folks. We need to provide them with software that is delivered as a service. We should stop selling them yet another server that will work as a band aid for just couple more years. Instead, provide them with elastic unlimited capacity. Provide them with software that will be upgraded by the Devops team that developed the software in the first place. Give them SaaS, give them public cloud. Make them efficient and they will show the world that they have in them to be leaders again. These organizations have smart business folks that are put in a constrained box by IT systems that tech industry had delivered them in last 2 decades.

In their greed to preserve margins, software vendors will try to sell them on expensive on-prem software and get them again on a road of frustration. This will be a bad thing. We need to educate the IT folks that buying on-prem software is not a smart thing to do and they are harming their organization and end users that work there.  Why? Because, on-prem software will always be inefficient to maintain, upgrade, scale properly and keep it in sync with advances in software.

Secondly, no matter what you do, do not tell these IT folks that they should do private cloud – do not do a smoke and mirrors show and try to sell same old crappy on-prem solutions as Cloud. Do not do it.

We created the problem of crappy on-prem software and systems in the first place; it is our responsibility to solve it the right way using a public cloud that is designed for long term, scale and maintenance. It doesn’t have to be AWS, it doesn’t have to GCP, it can be Verizon, it can be CenturyLink, it can be SoftLayer, it can be Profitbricks, but it cannot be same old software recompiled and packaged and relabeled ‘cloud’. I am not unrealistic to expect this to happen overnight, but if every replacement when needed going forward is with public cloud and SaaS, with interim Interop between on-prem systems and public cloud using something like Eucalyptus, that is a good journey for these organizations.

Now I have nothing against selling on-prem private cloud to Fortune 50 – they have  money and I view that as distribution of wealth. Heck, sell them few mainframes and 8″ tapes or even a 10000 pack floppy disks. Who cares, they got money to spend to keep their budgets growing.

Agree/Disagree/Have a Comment? Share your experiences from the real world. Tweet away or comment below.



9 thoughts on “A letter from the real world

  1. While I agree, as an industry, I would hope we can do even better.
    Because what about data security ? (if I’m not gonna ask it, some customers will)
    That is one of the other arguments for on-prem.
    For data security we need encryption and key management. Especially that last one seems to be hard to get right, One of the problems is, we’d need a solution for the whole stack. So far, I’ve seen very little.

  2. “Data is probably less secure on-prem anyways.”
    Yes, currently it is but only if the data is encrypted at rest.
    As I mentioned, I hope our industry could do even better.
    I was thinking about things like MIT Mylar which take a look at the whole stack.

  3. The real question is:
    Are customers willing to pay for rewriting their (sometimes custom) applications or change their processes to fit new applications ? Or will they choose a band-aid and use VDI on the public cloud. Because that is what I’m seeing happening in the short run. Instead of solving the real problem and finding or building a new application, they want to keep using their old application.

  4. “Secondly, no matter what you do, do not tell these IT folks that they should do private cloud – do not do a smoke and mirrors show and try to sell same old crappy on-prem solutions as Cloud. Do not do it.”

    Yet another salesman! What a bunch of marketing!

    The only way to build “the cloud” to be half-reliable and do what it is supposed to do is to build in-house clusters.

    Phase 1: build the complete, in house software stack, so that noone can screw with you;
    Phase 2: build a failover cluster;
    Phase 3: build multiple failover clusters;
    Phase4: start software development (yes, development, I uttered the taboo word!) and start building HA applications on top of HA clusters.

    *THAT* is “the could”. “The cloud” is not “software as a service”, nor is it a bunch of poorly managed, crappy, I/O constrained VM’s with hacked-up Linux images on them. Software as a service: how many buzzwords can you squeeze in?


    • Absolutely. I build my own servers! Joyent does the same, or rather, they dictate through a set of detailed specifications, how a server should be built, with which components, down to every single BIOS setting.

      Enough is enough!

      • Companies like Google, Facebook and Rackspace go even further. They order their parts directly from Asia and are involved in the design.

        Designing the right architectures for the task they need to perform is really important.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s