You are a CIO, a VP of Operations, or an IT manager. You could also be an Infrastructure Architect. You have been purchasing or using VMware products since it emerged in the early 2000s. You have lived and thrived through GSX, ESX, ESXi and beyond. You may also be the rarer breed that decided to use those free HyperV licenses and then realize it’s not free after all! Either way, you have faithfully kept your software up-to-date, spent months planning for upgrades and made your boss happy by generating those data-center productivity reports.
While you were doing this, your developers decided to take advantage of that corporate credit card and go to AWS — the cloud native world beckoned! Those cloud instances were easy to create — no need to call you while you were still setting up that vRealize Automation Suite! Running web apps was a breeze on AWS, so was Chef, Puppet, Mongo, Redis, Cassandra, and Hadoop. Everything was prefixed with the word “Elastic” and offered as a “Service” afterall.
Around that time you read a note from your favorite Industry Analyst (from Gartner, IDC, Forrester, take your pick) asking about your AWS strategy? As an IT organization did you not have an AWS strategy? Clearly you were doing something wrong, they alleged. Heeding their call you decided to make AWS mainstream, with a splattering of Azure. Next time you encountered your favorite Industry Analyst, you described your ambidextrous “dual cloud strategy” with a lot of enthusiasm. You were using both VMware and AWS.
Things were moving along until you realized you had a new set of problems. Your developers were spinning up multiple VMs only to find the top 5 best performing VMs to deploy their app — those noisy neighbors were stealing CPU cycles.
You realized that your costs hadn’t really gone down. Your users spun up VMs and forget to shut them down. Even worse, your data scientists needed to run Spark on XXL VMs which cost an arm and a leg.
You are at a crossroad now. Your dual cloud strategy looked more like dual leak strategy — money leaking on both sides.
Then you hear of OpenStack.
That elusive API driven cloud framework that incumbents could not build. It has all the promise — a free hypervisor, a free management stack — no more hypervisor tax, no more per hour billing that adds up to hundreds of thousands of dollars every month.
But there is no such thing as a free lunch, you wonder! One has to pay something to somebody to keep the wheels of commerce moving! At around the same time you get another missive from your Analyst — asking if your Private Cloud strategy is BiModal. OpenStack would help you become BiModal, it said. This time you decide to take the initiative to figure out what that means before others do.
OpenStack, as it turns out has been around for several years now. While you were still implementing your dual cloud strategy an army of geeks from all over the world were busy getting it to a mature state. It’s being used by companies big and small. One can use OpenStack to run both VMs and Containers — so you don’t have to miss out on all the Docker action.
Whether or not you choose to use OpenStack is for you to decide. The folks who decided to go with Linux over Solaris, HPUX or AIX went through a similar decision making process. When you do make that choice you also need to make a second choice — what kind of OpenStack based private cloud do you want?
It helps to realize that software systems, particularly distributed systems, have evolved significantly over the last decade. It is possible to build systems that can tolerate failure of hardware and software components — this reduces the administrative burden (read OPEX) on IT significantly. However, enterprise management systems have been bogged down by status quo in design. OpenStack services are designed to live in the new cloud native world — not controlled by traditional design paradigms. Ask your vendor how they have designed their OpenStack deployment. Is it a bunch of Ansible or Chef scripts wrapped in a jazzy UI? Does the deployment work only when an army of professional services team show up with their hammer and tongs? Or is the system designed to take advantage of Openstack’s service oriented architecture and has all of the bells and whistles that go into the making of a real cloud.
Here’s to OpenStack until the next strategic advice from your favorite Analyst!